European Community’s Framework Programme 6
EUROPEAN EXTREMELY LARGE
TELESCOPE INSTRUMENT DESIGN STUDY
MIDIR
The MID-InfraRed Instrument for the E-ELT
Document title:
Conceptual Design Study of MIDIR
Document number:
ELT-TRE-LEI-11200-0001
Issue No
1.0
Date
14 July 2006
Prepared by
Frank Molster (editor)
Approved by
Bernhard Brandl
Rainer Lenzen
Released by
Bernhard Brandl
Rainer Lenzen
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
CHANGE RECORD
Issue
Date
Section / Paragraph affected
Reason / Initiation / Remarks
0.0
15-3-2006
All
Starting document, only the TOC
0.1
20-3-2006
All
Update of the TOC
0.11
21-4-2006
2.1; 4.1; 4.4; 4.5; 5.1; 5.2; 5.4;
6.3; 7.3; 7.5; 7.6; 8.4;
Input from R. Lenzen & F. Molster
0.12
27-4-2006
4.2; 5.3; 5.4
Input from R. Stuik
0.13
1-5-2006
4.1; 6.6
Input from L. Venema
0.14
8-5-2006
2.1; 2.3; 2.5; 4.3; 5.3; 6.1; 6.2;
6.8; Annex B
Input from U. Kaufl, R. Stuik, P.
Hallibert & A. Glasse
Included reference document
0.15
9-5-2006
Ch3; 4.1; 4.5; 6.2; 6.9; 6.10
Input from L. Venema, A. Glasse & B.
Brandl
0.16
10-5-2006
Ch1;4; 7; numerous editorials
Input from B. Brandl
0.20
21-6-2006
All
Input from everybody
0.21
23-6-2006
All
Input from L. Venema, R. Lenzen & R.
Stuik
0.22
3-7-2006
Ch3; 9.2; 9.3, 9.4, 6.7
Input from L. Venema, B. Brandl, G.
Finger
0.30
5-7-2006
All
Input from all
0.31
11-7-2006
Ch 4,5,6
Input from B. Brandl, L. Venema
0.9
12-7-2006
Ch 5
Input from R. Lenzen, R. Stuik & B.
Brandl
1.0
14-7-2006
All, Ch7, Ch9
All kinds of editorial changes
Page 2 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table of Contents
1
EXECUTIVE SUMMARY...................................................................................................................... 6
2
BACKGROUND INFORMATION ....................................................................................................... 9
2.1
2.2
2.3
2.4
2.5
3
MAIN SCIENCE DRIVERS................................................................................................................. 15
3.1
3.2
3.3
3.4
3.4.1
3.4.2
3.5
3.6
3.7
3.8
3.9
3.9.1
3.9.2
3.9.3
3.10
4
STUDY TEAM COMPOSITION ............................................................................................................. 9
APPLICABLE DOCUMENTS ............................................................................................................... 10
REFERENCE DOCUMENTS ................................................................................................................ 10
BOUNDARY CONDITIONS & ASSUMPTIONS ..................................................................................... 11
ACRONYMS ...................................................................................................................................... 12
OVERVIEW ....................................................................................................................................... 15
CONDITIONS IN THE EARLY SOLAR SYSTEM ................................................................................... 15
DETECTION AND CHARACTERIZATION OF EXTRASOLAR PLANETS................................................. 16
FORMATION AND EVOLUTION OF PROTO-PLANETARY DISKS AND PLANETS ................................. 18
Spatial Signatures of Planet Formation.................................................................................... 20
Spectral Signatures of Disk Evolution....................................................................................... 21
THE GALACTIC CENTER .................................................................................................................. 26
THE LUMINOUS CENTERS OF NEARBY GALAXIES .......................................................................... 27
AGN AT HIGH REDSHIFTS............................................................................................................... 30
GAMMA-RAY BURSTS AT HIGH REDSHIFTS .................................................................................... 31
POLARIMETRY ................................................................................................................................. 32
Magnetic Fields in Star Formation ........................................................................................... 33
The Structure of Young Stellar Disks ........................................................................................ 33
The Geometry of Active Galactic Nuclei (AGN) ....................................................................... 33
SUMMARY........................................................................................................................................ 35
GENERAL CONSIDERATIONS ........................................................................................................ 36
4.1
4.2
4.3
4.3.1
4.3.2
4.4
4.4.1
4.4.2
4.5
4.5.1
4.5.2
4.5.3
4.6
4.6.1
4.6.2
4.6.3
4.7
4.8
4.8.1
4.8.2
4.8.3
4.8.4
4.8.5
4.8.6
4.8.7
4.8.8
TOP LEVEL REQUIREMENTS FOR MIDIR ........................................................................................ 36
CONSIDERATIONS ON DIFFRACTION LIMITED PERFORMANCE ......................................................... 36
THERMAL BACKGROUND ................................................................................................................ 39
Why Chopping?.......................................................................................................................... 39
Some Background Information.................................................................................................. 43
REQUIREMENTS FOR THE IMAGING AND LOW RESOLUTION SPECTROSCOPY MODE ...................... 46
Imaging scale ............................................................................................................................. 46
Filter Selection........................................................................................................................... 47
REQUIREMENTS FOR THE MEDIUM AND HIGH RESOLUTION SPECTROMETER ................................ 49
General Considerations............................................................................................................. 50
Requirements for the Medium Resolution Spectrometer .......................................................... 55
Requirements for the High Resolution Spectrometer................................................................ 58
CONSIDERATIONS FOR POLARIMETRY............................................................................................. 59
Introduction: .............................................................................................................................. 59
Persistent Speckles..................................................................................................................... 61
Design Considerations............................................................................................................... 61
DATA RATES.................................................................................................................................... 62
CALIBRATION: REQUIREMENTS AND SOLUTIONS ........................................................................... 64
Introduction................................................................................................................................ 64
Variability of the Sky.................................................................................................................. 65
Telescope Thermal Background ................................................................................................ 66
Detector Variations and non-Linearity ..................................................................................... 66
Variability in Telescope and Instrument ................................................................................... 66
Instrument Characterisation/Calibration ................................................................................. 66
Calibration Hardware Components.......................................................................................... 67
Calibration Strategy .................................................................................................................. 68
Page 3 of 204
Conceptual Design Study of MIDIR
5
ELT-TRE-LEI-11200-0001
1.0
ATMOSPHERIC EFFECTS AND ADAPTIVE OPTICS ................................................................ 70
5.1
5.2
5.3
5.4
5.5
6
Doc. No
Issue
ATMOSPHERIC DISPERSION ............................................................................................................. 72
ATMOSPHERIC TURBULENCE .......................................................................................................... 74
TIME DEPENDENT CHROMATIC EFFECTS ........................................................................................ 75
ATMOSPHERIC WATER VAPOUR ..................................................................................................... 76
AO REQUIREMENTS AND PERFORMANCE ....................................................................................... 78
CONCEPTUAL DESIGN...................................................................................................................... 82
6.1
ADAPTIVE OPTICS RELAY OPTICAL DESIGN .................................................................................... 82
6.1.1
AO relay preliminary specifications.......................................................................................... 82
6.1.2
Preliminary Optical Concept for a F/4.5-F/10 relay................................................................ 82
6.1.3
Preliminary Optical Concept for a F/16-F/10 relay................................................................. 85
6.1.4
Considerations on the Impact of Chopping on the Optical Design.......................................... 87
6.1.5
AO Relay Optical Design: Conclusion...................................................................................... 88
6.2
OPTICAL: DESIGN PRE-OPTICS ......................................................................................................... 88
6.2.1
Pre-optics: Common path.......................................................................................................... 88
6.2.2
Spectrometer Collimator ........................................................................................................... 91
6.2.3
The Spectrometer Pre-Optics .................................................................................................... 93
6.2.4
The Dichroic Chain ................................................................................................................... 94
6.2.5
The Integral Field Unit (IFU) ................................................................................................... 95
6.3
OPTICAL: DESIGN IMAGER ............................................................................................................... 98
6.3.1
The Collimator ........................................................................................................................... 98
6.3.2
The TIR-Camera ........................................................................................................................ 99
6.3.3
The MIR-Camera ..................................................................................................................... 100
6.3.4
The Grisms ............................................................................................................................... 101
6.4
OPTICAL: DESIGN HIGH RESOLUTION SPECTROMETER ................................................................ 102
6.4.1
Introduction.............................................................................................................................. 102
6.4.2
Global optical design............................................................................................................... 103
6.4.3
N-band system in detail ........................................................................................................... 105
6.5
OPTICAL: DESIGN MEDIUM RESOLUTION SPECTROMETER ............................................................. 111
6.6
OPTICAL DESIGN: CALIBRATION UNIT ........................................................................................... 112
6.7
DETECTORS AND FOCAL PLANE CONFIGURATIONS ....................................................................... 114
6.7.1
2K x 2K λc=5 µm Hawaii-2RG arrays .................................................................................... 114
6.7.2
1K x1K Si:As Aquarius Arrays ................................................................................................ 119
6.7.3
Infrared Wavefront Sensor ...................................................................................................... 122
6.8
CHOPPING ...................................................................................................................................... 123
6.8.1
Technical Alternatives ............................................................................................................. 123
6.8.2
Trade-Offs ................................................................................................................................ 125
6.8.3
Recommendations and Suggestions for Prototyping .............................................................. 125
6.9
CRYOSTAT CONCEPT AND TEMPERATURE REQUIREMENTS.......................................................... 126
6.9.1
Temperature Requirements...................................................................................................... 126
6.9.2
Cooling Schemes...................................................................................................................... 126
6.9.3
Background Information.......................................................................................................... 136
6.10
MECHANICAL SETUP AND METROLOGY SYSTEM ......................................................................... 143
6.10.1
General Considerations...................................................................................................... 143
6.10.2
The Baseline Mechanical Design ....................................................................................... 148
6.10.3
Size and Mass Estimates ..................................................................................................... 152
7
INSTRUMENT SENSITIVITIES AND COMPARISONS............................................................. 153
7.1
7.2
7.3
7.3.1
7.3.2
7.4
7.5
7.5.1
ASSUMPTIONS AND CALCULATIONS ............................................................................................. 153
IMAGER SENSITIVITY..................................................................................................................... 157
SPECTROGRAPH SENSITIVITY ........................................................................................................ 157
Performance of the R=3000 medium resolution spectrograph .............................................. 157
Performance of the R=50,000 (25,000) High Resolution Spectrograph................................ 161
EXTENDED SOURCE SENSITIVITY .................................................................................................. 163
OTHER MID-IR FACILITIES (CURRENT AND FUTURE) .................................................................. 163
Mid Infrared Instrumentation on 8m-class Telescopes .......................................................... 164
Page 4 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
7.5.2
Expected Contemporaries of MIDIR ....................................................................................... 165
7.6
PERFORMANCE COMPARISONS ...................................................................................................... 169
8
SPECIFIC MIDIR REQUIREMENTS ON THE TELESCOPE ................................................... 173
8.1
8.2
8.3
8.4
8.4.1
8.4.2
8.4.3
8.4.4
8.5
9
REQUIREMENTS ON THE TELESCOPE SITE ..................................................................................... 173
REQUIREMENTS ON THE TELESCOPE FOCUS ................................................................................. 173
REQUIREMENTS ON THE TELESCOPE PERFORMANCE .................................................................... 176
FIELD- AND PUPIL ROTATION ........................................................................................................ 177
Instrumental De-rotation......................................................................................................... 177
Detector De-rotation ............................................................................................................... 177
Optical De-rotation.................................................................................................................. 177
De-rotation by Post-processing............................................................................................... 178
SUITABILITY OF MIDIR AS A “FIRST LIGHT” INSTRUMENT ......................................................... 179
MANAGEMENT.................................................................................................................................. 181
9.1
BUDGET ......................................................................................................................................... 182
9.1.1
Introduction.............................................................................................................................. 182
9.2
TIME LINE ...................................................................................................................................... 185
9.3
BASELINE OVERVIEW AND RISK ITEMS ........................................................................................ 186
10
CONCLUSIONS AND OUTLOOK................................................................................................... 192
ANNEX A: NOETHE 2003 ........................................................................................................................... 194
ANNEX B: R. SIEBENMORGEN & H.U. KÄUFL 2006 ......................................................................... 196
Page 5 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
1 Executive Summary
MIDIR is a combined imager and spectrograph for the European Extremely Large
Telescope, E-ELT. It will cover the wavelength range from 3 to 20μm with a goal to
extend the wavelength coverage to 27μm if the atmospheric properties of the site are
sufficiently good. Because of the naturally high thermal background from telescope and
atmosphere the main applications for MIDIR will be imaging and spectroscopy at highest
angular resolution and high spectral resolution. In these areas MIDIR will be
complementary or even superior to future space facilities like JWST-MIRI. Additional
capabilities of MIDIR include quick response times to targets of opportunity and high time
resolution (order of milli-seconds). To reach its maximum resolution and sensitivity,
MIDIR will require an adaptive optics (AO) system. Due to the thermal emission from
additional warm surfaces in the optical train MIDIR requires an IR-optimized and cooled
AO system.
The combination of an E-ELT at a good site with a dedicated mid-IR instrument enables
compelling science cases in numerous areas from the conditions in the early Solar system
to Gamma-ray bursts at very high redshift. Including the characterization of exoplanets,
the formation and evolution of proto-planetary disks and the luminous centers of active
galaxies MIDIR is best suited to study the origins of life in the Universe and the evolution
of galaxies. Because of the instrument’s flexibility, the discovery space of MIDIR does
not crucially depend on the projection of current science “killer applications” 15 years into
the future.
MIDIR is one of eight instruments currently being studied for the E-ELT. This report
summarizes the results from a nine months long instrument “Small Study”, which has
been partially funded by the EU. The work within this study has been structured as
follows: for a given telescope and wavelength regime we have composed a unique suite of
compelling science cases. These science cases were translated into top level instrument
requirements, which were broken down into a series of technical specifications, yielding
the basis for the baseline instrument design. Trade-off studies between possible design
options were included where necessary and possible. At this point, instrument cost and
complexity have not been considered as critical boundary conditions.
Table 1-1 lists the main instrument/AO parameters and requirements on the telescope. Our
study shows that a first-rate mid-IR instrument on the E-ELT is scientifically
recommended and technically feasible. The guaranteed scientific return and the reduced
demands on the wavefront quality (with respect to optical/near-IR instruments) suggest
MIDIR as a first-light E-ELT instrument. MIDIR does not require developments of
fundamentally new technologies, but extends certain technologies beyond the current
state-of-art. However, several issues need to be addressed in more detail in future studies.
Page 6 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 1-1 Summary of instrument, telescope and AO requirements and parameters.
Parameter
Value
Instrument parameters
Wavelength range
Instrument modes
Field of view
3.5 – 20μm (goal: 3.5 – 27μm)
•
broad/narrow-band imaging
•
low resolution, long slit spectroscopy (R~300)
•
medium resolution IFU spectroscopy (R~3000)
•
high resolution IFU spectroscopy (R~50,000)
~ 40″ × 40″ (imaging)
~1″ × 1″ (IFU spectroscopy)
Image quality
diffraction limited at all wavelengths and field positions
Entrance window
~150 – 250 mm ∅
Mass
4700 kg (incl. electronics)
Size
3 × 2.3 m3 + 1.4 m3 (without AO)
Telescope requirements
Acceptable telescope f/#
4.5 – 15
Minimum scientific field size
1.5' × 1.5'
Straylight baffling
no warm baffles
Thermal emission
optimized for low thermal background and minimum number of
surfaces
Maximum zenith angle
60 degrees (limited by AO performance)
Focal station
Cassegrain or Nasmyth (see Section 8.2)
Back focal distance
≥ 500 mm
Instrument attachment
off-line image de-rotation in software, fixed pupil
Chopping
no requirements
Pointing/tracking accuracy
~1″ (1-σ)
Telescope site
as high (h ≥ 4000m) and dry (PWV << 1mm) as possible
AO requirements
Principle
single-conjugate system, specific to MIDIR
Operation
encapsulated and cooled to TBD Kelvin (mid-IR optimized)
Performance
≥ 50% SR at L&M, 80% SR at N&Q
Correctable FOV
≥ 40″ × 40″
ADC
intern, if required at all (TBD)
Page 7 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
This document is organized as follows: First we presents the science case for MIDIR from
which the top level requirements are derived, followed by general considerations on
diffraction-limited performance, chopping, and the requirements for the various
instrument modules (imager, low-, medium-, and high-resolution spectrograph).
Atmospheric properties (transmission, emission, dispersion and turbulence) are being
discussed next. Chapter 6 is the main part providing conceptual designs for the AO
system, the pre-optics, the imager and the spectrograph modules main optics, as well as
considerations for calibration, detectors, chopping techniques, cryostat concepts and
mechanical setup. Then we estimate the sensitivity of MIDIR and compare it to other
facilities, followed by a discussion of MIDIR-specific requirements on telescope and site.
The study document concludes with a budget estimate, project schedule, a list of risk
items, and an outlook beyond this study.
Page 8 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
2 Background Information
2.1 STUDY TEAM COMPOSITION
MIDIR is a joint project of five institutes with the following core team members:
Leiden University:
Bernhard Brandl
Frank Molster
Remko Stuik
(PI)
(Project coordinator)
ASTRON:
Lars Venema
Ton Schoenmaker
(Project engineer)
Rainer Lenzen
Wolfgang Brandner
(Co-PI)
MPIA:
ESO:
Gert Finger
Ulli Käufl
UK ATC:
Alistair Glasse
David Lee
Besides the above mentioned people, this study would not have been possible without the
valuable input from the following people (in alphabetical order):
Hermann Böhnhard (MPS)
Raymond van den Brink (ASTRON)
Benedetta Ciardi (MPA)
Ewine van Dishoeck (Leiden Univ.)
Wolfgang Gässler (MPIA)
Miwa Goto (MPIA)
Pascal Hallibert (Leiden Univ.)
Christoph Keller (Utrecht Univ.)
Dietrich Lemke (MPIA)
Miska Le Louarn (ESO)
Jan Noordam (ASTRON)
Chris Packman (Univ. of Florida)
Jan-Willem Pel (ASTRON)
Johan Pragt (ASTRON)
Almudena Prieto (MPIA)
Ronald Roelfsma (ASTRON)
Ralf Siebenmorgen (ESO)
Daphne Stam (UvA)
Michael Sterzik (ESO)
Jaap Tinbergen (ASTRON)
Paul van der Werf (Leiden Univ.)
Ralph Wijers (Univ. of Amsterdam)
Sebastian Wolf (MPIA)
Page 9 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
2.2 APPLICABLE DOCUMENTS
No applicable documents have been identified.
2.3 REFERENCE DOCUMENTS
The following documents are for reference purposes only:
Table 2-1: Reference documents
RD Title
Author
Version (date)
1
OWL Instrument Concept R. Lenzen &
Study
T-OWL,
Thermal B. Brandl
Infrared Instrument for OWL.
Doc nr: OWL-CSR-ESO00000-0161
2
A sky-noise measurement and
its implication for groundbased infrared astronomy in the
10-micron atmospheric
window.
3
Observation capabilities and L. Venema et
technical solutions to a thermal al.
and MIR instrument for ELTs
4a
MIDIR/TOWL: the
thermal/mid-IR instrument for
the E-ELT.
Observational capabilities and
technical solution of a thermal
and MIR instrument for ELTs.
B. Brandl et
al.
5
Observing extended objects
with chopping restrictions on
8m class telescopes in the
thermal infrared
H. U. Käufl
6
M. Bertero et
Wide-Field Imaging at Midal.
Infrared Wavelengths:
Reconstruction of Chopped and
Nodded Data
PASP 112, Issue 774, p. 11211137 (2000)
7
Robust reconstruction from
chopped and nodded images,
F. Lenzen, O.
Scherzer, S.
Schindler
A & A 443, Issue 3, December I
p.1087-1093 (2005)
8
Effects of Atmospheric Water
Colavita et al. PASP 116, p.876-885 (2004)
4b
H. U. Käufl
et al
1.1 (05 Oct 2005)
Exp. Astron. 2, 115-122 (1991)
Visions for IR Astronomy
Proceedings, Paris, March 2006
SPIE proceedings 6269-75
SPIE proceedings 6269-186,
Orlando (2006)
R. Lenzen et
al.
ESO Conf. & Workshop Proc.,
Proc. of an ESO / ST-ECF
workshop on calibrating and
understanding HST and ESO
instr. (1995)
Page 10 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Vapor on Infrared
Interferometry
9
Adaptive Optics for
Astronomical Telescopes
Hardy
Oxford University Press (1998)
10
Estimation of thermal
conduction loads for structural
supports of cryogenic
spacecraft assemblies
Ronal G.
Ross, Jr
Cryogenics 44, p.421-424 (2004)
11
VISIR, the mid-infrared imager
and spectrometer for the VLT
Y. Rio et al.
SPIE Conf Proc. 3354-1, p.615
(1998)
12
Design for the 5-28 µm NGST
MIRI spectroscopy channel
M. Wells et al SPIE Conf. Proc. 4850, p.504
(2003)
13
MIRI spectrometer optical
design
B. Kruizinga
et al.
14
CanariCam-Polarimetry: A
Dual-Beam 10 mum
Polarimeter for the GTC
Packham et
al.
15
Infrared helioseismology Detection of the chromospheric
mode
D. Deming et
al
Nature, vol. 322, p.232-234
(1986)
16
Infrared Heterodyne
Spectroscopy - a Tool for
Helioseismology
DA Glenar et
al.
Adv. Helio- and
Asteroseismology: IAU symp.
123, p.481 (1988)
Proceedings of ‘Fifth
International Conference on
Space Optics’
2.4 BOUNDARY CONDITIONS & ASSUMPTIONS
Throughout this document we have made the following assumptions:
•
An ELT with a primary mirror diameter of 42m (baseline), and a possible range from
30 – 60m.
•
A high and dry telescope site like Chajnantor (unless mentioned otherwise)
•
A median optical seeing of 0.8″
•
No financial budget limit
•
First light ~2015
Page 11 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
2.5 ACRONYMS
5M
Five Mirror (design)
ADC
Atmospheric Dispersion Compensator
AGN
Active Galactic Nuclea
AO
Adaptive Optics
AR
Anti-Reflectivity
AU
Astronomical Unit
BB
Black Body
BIB
Blocked Impurity Band
BLIP
Background Noise Limited Performance
CONICA
Coude Near Infrared Camera
CS
Colour Sensitive sensors
CTE
Coefficient of Thermal Expansion
DE
Dispersive Element
DIT
Detector Integration Time
DL
Diffraction Limited
DM
Deformable Mirror
EC
European Commission
(E-)ELT
(European) Extremely Large Telescope
ESE-WG
ELT Science & Engineering Working Group
ESO
European Southern Observatory
FOV
Field of View
FP
Fabry-Perot
FPA
Focal Plane Array
FPM
Focal Plane Module
FP6
Framework Programme 6
FT
Fourier Transform
GM
Gifford McMahon (cooler)
HR
High Resolution
IFU
Integral Field Unit
IRAC
Infrared Array Camera
IRTF
Infrared Telescope Facility
IS
Integrating Sphere
Page 12 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
ISAAC
Infrared Spectrometer and Array Camera
JWST
James Webb Space Telescope
KMOS
K-band Multi-Object Spectrograph
LGS
Laser Guide Star
LTAO
Laser Tomographic Adaptive Optics
Mag
magnitude
mas
milli-arcsecond
MHR
Medium/High Resolution
MICHELLE
Mid-infrared (7-26 micron) imager and spectrometer (Gemini)
MIDI
Mid-IR Interferometric Instrument for VLTI
MIDIR
Mid IR instrument
MIRI
Mid-infrared Instrument (on the James Webb telescope)
MLI
Multi Layer Insulation
MLOF
Mount Lemmon Observing Facility
MR
Medium Resolution
MTBF
Mean Time Between Failure
N/A
Not applicable
NGS
Natural Guide Star
OPD
Optical Path Difference
OWL
Overwhelmingly large telescope (100m)
PAH
Polycyclic Aromatic Hydrocarbon
PDS
Point Design Study
PS
Point Source
PSF
Point Spread Function
PTC
Pulse Tube Cooler
PWV
Precipetable Water Vapour
R
Resolution power (λ/Δλ)
RC
Richey-Chrétien (design)
SCAO
Single Conjugate Adaptive Optics
SMBH
Super-Massive Black Hole
TBC
To Be Confirmed
TBD
To Be Determined
TIMMI
Thermal Infrared MultiMode Instrument
TMA
Three Mirror Anastigmat
Page 13 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
TMC
Tuneable Monochromator
TMT
Thirty Metre Telescope
TNTCAM
Ten and Twenty micron mid-IR array Camera.
TS
Telescope Simulator
VISIR
VLT Imager and Spectrometer for mid Infrared
VLT
Very Large Telescope
WBS
Work Breakdown Structure
WIRO
Wyoming Infrared Observatory
WP
Work package
YSO
Young Stellar Object
ELT-TRE-LEI-11200-0001
1.0
Page 14 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
3 Main Science Drivers
3.1 OVERVIEW
The combination of a large 30 – 60m telescope aperture with a good telescope site, and
low-noise, large-format mid-IR array detectors will open up completely new perspectives
for mid-IR astronomy from the ground, way beyond the areas of classical mid-IR
astronomy. The high point source sensitivity will allow studies of objects at very high
redshifts, and high-resolution spectroscopy can be used for morphological and kinematical
studies in unsurpassed details.
In general, cooled space based observatories are considerably more sensitive in the mid-IR
to faint surface brightness objects than ground-based observatories. Space-based
observatories, however, because of their restricted aperture size (85cm for Spitzer, up to
6.5m for JWST), are rather limited in terms of angular resolution compared to a groundbased ELT. For example, at 10 µm JWST has a spatial resolution of 0.35” whereas the
diffraction limit for a 42m ELT is 0.045'', corresponding to only a few AU in the nearest
star-forming regions, 200 AU at the Galactic Center and x pc at the Virgo cluster.
Consequently, most MIDIR science cases from the ground focus on high angular
resolution and compact objects, rather than the study of faint surface brightness features.
Moreover, space-based missions lack high spectral resolution instruments. Specifically,
exciting science cases for MIDIR focus on:
•
highest angular resolution
•
very high spectral resolution
•
quick response times (< 1 day)
•
time variability (in the order of milli-seconds to minutes)
In the following sections we discuss several example science cases. However, it is
important to note that the MIDIR capabilities are by no means limited to these topics.
3.2 CONDITIONS IN THE EARLY SOLAR SYSTEM
Comets are considered to represent the most primordial bodies in our solar system
accessible to Earth-based observations to date. The structure and composition of cometary
ices are key to understanding the formation and evolution of matter within the early solar
system (Bockelée-Morvan et al. 2004). Ices are particularly sensitive to temperature and
radiation processing. Comets likely formed at diverse distances from the sun and outside
the ‘frost line’ in the solar nebula (~5AU) out to the Kuiper Belt (40-50AU). Depending
on the temperature – thus the distance – of the formation region, dust alteration from
amorphous to crystalline forms may have occurred. Thus, the composition of a comet in
its icy and dusty components contains the signatures from the formation period. Today,
icy and dusty planetesimals primarily reside in two reservoirs: the Oort cloud contains the
long-periodic and ‘dynamically new’ comets, the Kuiper Belt is the main source for the
short-periodic comets. A third reservoir inside the 'frost line', the Main Asteroid Belt, has
Page 15 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
recently been discovered and their existence lends support to theories that these could be a
major source of the water on Earth (Hsieh & Jewitt 2006).
Sensitive mid-IR spectroscopy provides the most important tool to understand the
“weathering” cometary nuclei have undergone since their formation. In principle, the spin
statistics of protons in H2O and NH3 in the gaseous outflows are considered a reliable
thermometer to probe the temperature history over the past 106 years – but only if the spin
statistics have not been altered once the molecules were released from the comet nucleus.
The high spatial resolution which can be achieved with MIDIR will probe the unaltered
gas within seconds after it has left the surface.
The richest wavelength domain for volatile studies is the thermal IR between 3 and 5μm,
and for the dust the 7 to 23μm range. Using high-dispersion spectroscopy in the former
case a number of parent gas species from cometary ices (H2O, CO, NH3, CH4, C2H2, C2H6,
CH3OH, HCN) can be measured. The constitutional structure and composition of the dust
is revealed by low and medium resolution spectroscopy. Both techniques have been used
at 8m-class telescopes, but their application is limited to the very brightest comets. For
illustration, Figure 3-3 shows the time-averaged, flux-calibrated spectra for comet Tempel
1 taken from a pencil beam centered on the comet nucleus (280 x 1109 km). Clearly
detected are about 16 spectral lines of H2O (panels A,C,D), six Q-branches of C2H6 along
with features of CH3OH (panels C-F), and eight spectral lines of HCN, along with two
lines of C2H2 (panel E). However, a more comprehensive map of the formation regions
and conditions in the early solar system can only be obtained from 30-50m-class
telescopes.
3.3 DETECTION AND CHARACTERIZATION OF EXTRASOLAR
PLANETS
To date more than 180 extrasolar planets have been detected1. On the vast majority of
these planets little information is available apart from a lower mass limit. Most of these
planets have been detected indirectly via their gravitational pull on the parent star causing
periodic Doppler shifts. However, direct detections of the light emitted from the planets
are needed to derive physical parameters such as temperature, chemical composition, and
atmospheric structure and composition. The knowledge of these parameters is crucial to
our understanding of the formation and evolution of the planets and for comparative
planetology, in particular to the planets in our Solar system. The ultimate goal, of course,
is to understand the presence and development of life elsewhere in the Universe.
Direct detections of extrasolar planets are extremely challenging because the planetary
radiation is intrinsically very weak and the angular separation between planet and its much
brighter parent star is very small. To detect the radiation from the planet two general
methods can be used:
A. Spatially resolving the planet from the star
B. Separating the radiation spectro-photometrically.
1
See Jean Schneider's Extrasolar Planet Encyclopaedia at http://exoplanet.eu.
Page 16 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Examples of spatially resolved extrasolar planets around brown dwarfs have been reported
by Chauvin et al. (2005) and Guenther et al. (2005). Here we will focus on method B, i.e.,
searching for planetary signatures in a spatially unresolved planet-star system.
The most effective approach is to select systems with planets that are known to transit
their star. Over the coming years, various missions, such as Kepler and Corot, will be
launched to search for transiting planets, and the number of known transiting planets is
expected to increase significantly.
Figure 3-1 Spectra of 2 MJ brown dwarfs at 0.1, 0.3, 1 and 5 Gyrs. The Spitzer and JWST
sensitivities are also plotted. Methane absorption features are strong over this full range of
ages, while the ammonia features strengthen with age (Figure from Burrows, Sudarsky &
Lunine 2003).
Figure 3-1 shows model spectra of a 2 MJ planet/brown dwarf to illustrate the strong
absorption features of methane and ammonia, which fall into the wavelength range
spectroscopically covered by MIDIR. These features are not present in the pure stellar
spectra and, depending on the planets position, will vary with time as the planet orbits the
stars. In other words, small relative changes in molecular absorption features between
spectra taken over an orbital period can be used to directly detect and better characterize
the planet.
A successful variation of this technique has been reported by Deming et al. (2005) and
Charbonneau et al. (2005), using the small change in broad-band flux density detected by
Spitzer as the planet orbits in front of the star. A comparison between these two
observations with theoretical models is shown in Figure 3-2. While the photometric
stability of Spitzer will not be achieved with a ground-based instrument two aspects are in
favour of MIDIR:
1. The planet-to-star flux ratio (i.e, the contrast) is most favorable in the N-band,
with up to two magnitudes of improvement over the near-IR (Figure 3-2)
2. The spectroscopic detection depends only on the relative changes of spectral
features and does not depend on the absolute photometric stability.
Page 17 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Hence, there is a good chance of characterizing exoplanets in a unique way with MIDIR,
but further modelling is needed to quantify the potential of this method.
Figure 3-2 The planet-to-star flux density ratio as a function of wavelength for models of
the transiting planets TrES-1 (purple) and HD~209458b (green). Superposed are the data
on TrES-1 from IRAC (gold) and on HD 209458b from MIPS 24μm (green). Also shown
are the band-averaged fluxes for the models in the four IRAC bands (TrES-1: yellow; HD
209458b: blue) (Figure from Burrows, Hubeny & Sudarsky 2005).
3.4 FORMATION AND EVOLUTION OF PROTO-PLANETARY DISKS
AND PLANETS
Circumstellar disks are a natural by-product of the star formation process. The material in
these disks comprises the building blocks for future planetary systems (e.g., Lissauer
1993, Beckwith et al. 2000). Virtually all of the planets detected to date are gaseous
Jupiter-like planets, which are thought to form in disks within 1–10 Myr after the
formation of the parent star (Pollack et al. 1996). In the core-accretion model, a few rocky
cores with masses of 10–20 Earth masses must have formed quickly enough to attract gas
to form a gas-rich planet. Over time, at most 20 Myr, the gas in the disk will dissipate and
the small grains will coagulate or be blown away. This then leads to the debris disk phase
in which the disk is optically thin at UV and IR wavelengths and the grains are of
secondary origin, replenished by collisions of larger objects: asteroid-sized bodies or
planetesimals. For example, Herbig Ae/Be stars (t = 1–5 Myr) are the immediate
progenitors of classical debris disks like Vega (A0V, t ~100 Myr), β-Pic (A5V, 20 Myr) or
Fomalhaut (A3V, 100 Myr). Rocky planets with masses comparable to those of the Moon
or Earth form by gradual accretion of these planetesimals.
Page 18 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
The transition phase from gas-rich to gas-poor disks is clearly a pivotal period in planet
formation, and direct observations of the gas content, kinematics and composition are the
key to constrain the processes and time-scales involved. In particular, the bulk of the disk
mass is in the gas, not in the dust. MIDIR's spatial and spectral resolution will be crucial.
For example, a maximum resolution of 27 milli-arcsec at 4.7 μm corresponds to a few AU
in proto-planetary disks around T Tauri stars in the nearest star-forming regions (Oph, Cha
or Lupus at 150 pc).
Figure 3-3 Detection of parent volatiles and dust in comet Tempel 1 after the impact
event. The dashed line in each panel represents the cometary continuum convolved with a
synthetic transmittance spectrum of the terrestrial atmosphere (Mumma et al. 2005).
Page 19 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
3.4.1 Spatial Signatures of Planet Formation
Proto-planetary disks have now been imaged from the optical/near-IR to millimeter
wavelengths around low-mass T Tauri stars (e.g. Dutrey et al. 1994, 1998), intermediatemass Herbig Ae/Be stars (Mannings & Sargent 1997, 2000), and possibly also around
massive stars (e.g., Schreyer et al. 2002, Fontani et al. 2004). Cleared, inner dust disk radii
have been measured for several stars: ~4 AU around the 10 Myr old TW Hydrae (Calvet et
al. 2002), ~10 AU around Coku Tau/4 (D'Alessio et al. 2005, Quillen et al. 2004), and ~4
AU around GM Aur (Rice et al. 2003). The gap in emission is characterized by a depletion
of, at least, the population of small dust grains, which are responsible for the near- to midIR fluxes. The direct confirmation (via imaging) of these indirectly (via SED modelling)
determined gaps will provide valuable constraints on the evolution of the planet-forming
region and thus on the process of planet formation itself.
Once (proto-)planets have been formed, they may significantly alter the surface density
profile of the disk and thus create signatures in the disk that are much easier to find than
the planets themselves (Figure 3-4). The appearance and type of these signatures depend
on the mass and orbit of the planet, but even more on the evolutionary stage of the
circumstellar disk. While the spatial structure of optically thick, young circumstellar disks
around Herbig Ae/Be and T Tauri stars is dominated by gas dynamics, the much lower
optical depth and gas-to-dust mass ratios in debris disks make the Poynting-Robertson
effect and stellar wind drag, in addition to gravitation, responsible for the resulting disk
density distribution (e.g., Zuckermann, Forveille, & Kastner 1995; Liseau & Artymowicz
1998).
Figure 3-4 Response of a gaseous disk to an embedded planet (lf 2001 with data from W.
Kley).
Hydrodynamic simulations of gaseous, viscous proto-planetary disks with an embedded
proto-planet show that the planet may open and maintain a significant gap (e.g., Bryden et
al. 1999; Kley 1999, 2000). This gap, which is located along the orbit of the planet, may
extend to several astronomical units in width. The gas accretion on the planet can continue
to planet masses up to ~10 MJupiter, where tidal forces become sufficiently strong to
prevent further gas flow into the gap. The simulations also show that only planets with
masses >0.1 MJupiter produce significant perturbations in the disk's surface density (Bate et
al. 2003). Paardekooper & Mellema (2004) found that for typical disk masses of 0.01 Mo
within 100 AU the strong spiral shocks near the planet are able to decouple the larger
particles (~0.1 mm) from the gas. This leads to the formation of an annular gap in the dust,
Page 20 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
even if there is no significant gap in the gas density. As the opacity at longer wavelengths
is dominated by these larger particles, the signatures of low-mass planets in disks can be
stronger than previously thought. For 1mm-sized particles the minimum mass for a planet
to open a gap this way was found to be 0.05 MJupiter – corresponding to ~16 MEarth. Figure
3-5 presents a simulation for better illustration. The star is assumed to be a Herbig Ae star
(T=10,000K, L=46Lo) at a distance of 60pc; the disk parameters are those of the Butterfly
star (Wolf, Padget, & Stapelfeldt 2003) assuming a flared, Shakura-Sunyaev-type disk
with a disk mass of 0.01Mo and an outer radius of 100AU. The dust grain size distribution
and chemical composition are those of typical interstellar medium dust.
The dominant observable quantity originating from the inner disk region (r ≤ 10–20AU) is
the emission of mid-IR continuum radiation by hot dust. Given the typical distance of
nearby star-forming regions and the high angular resolution achievable on an ELT, MIDIR
will pioneer the study of planet-forming regions in circumstellar disks. The instrumental
requirements for this project are 3 – 20μm diffraction-limited imaging over a several
arcsec field of view through wide and intermediate band filters. The simulations in Figure
3-5 show that gaps in the disks around intermediate mass stars at a distance of 60 pc can
be detected with MIDIR on a 42m ELT.
undisturbed disk
2AU hole
4AU hole
i=0deg
i=60deg
Figure 3-5 Simulations of the re-emitted light at 10 microns, assuming an undisturbed disk
and a disk with a hole with a radius of 2AU and 4 AU, respectively, convolved with a 42mELT point spread function. The region shown corresponds to 60 AU x 60 AU.
3.4.2 Spectral Signatures of Disk Evolution
The highest spectral resolution is needed to probe the gas kinematics of molecules, such as
CO. The following discussion lists the main diagnostics for studies of proto-planetary
disks. Note that many of these same features are also excellent diagnostics of more deeply
embedded protostars, outflows and normal interstellar material – other exciting science
cases which have not been included to keep the discussion short.
Page 21 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
•
Evolution of Dust Grains. The N-band contains the resonance of the Si-O stretching
mode, characteristic of silicate dust grains, which can be found in pre-main sequence
stars with circum-stellar disks and solar-system comets. The shape of this band is
particularly sensitive to grain growth – the first step in planet formation- and
crystallization – a diagnostic of thermal processing, mixing and/or planetesimal
destruction in the disk (Figure 3-6). Pioneering spatially resolved spectroscopy has
opened the completely new field of mineralogy as a function of position in disks (e.g.,
van Boekel et al. 2004a,b), but this has been possible for only a handful of sources.
MIDIR, unlike VLTI/MIDI, will be sensitive enough to observe a statistically relevant
sample of disks. Instantaneous N-band spectral coverage at low spectral resolution is
the preferred mode.
•
H2 pure rotational lines in the MIR. H2 is the main gaseous reservoir in disks, and an
essential ingredient for building gas giant planets. The pure-rotational mid-IR lines are
the lowest possible transitions to search for, but they are extremely difficult to detect
because they are intrinsically very weak and always superposed on a strong
continuum. For a typical sensitivity of 10-16erg s-1cm-2 at 17µm at R=50000 (10σ, 1hr),
MIDIR can detect ~10 MEarth of H2 gas in a disk at 150 pc. For a disk at the distance of
the TW Hya association (56 pc), models predict S(1) and S(2) fluxes around 2x1015
erg s-1cm-2 if the star has excess UV radiation (Nomura & Millar 2005), readily
detectable with MIDIR. A MIDIR key program would be a survey in the S(1) 17µm
and S(2) 12µm lines in a large set of disks of various evolutionary stages, taken from
samples defined, e.g., by Spitzer. An integral field unit would be crucial for these
studies to correct for contaminating cloud emission. MIDIR’s spatial resolution of
0.05-0.09'' is well matched to the H2 emitting region of disks (10 AU = 0.07" at typical
distances of 150 pc for T Tauri stars in Oph, Cha, or Lupus).
Page 22 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 3-6 Continuum-subtracted 10μm silicate emission bands observed toward a
selected sample of isolated Herbig Ae stars with ISO-SWS. The silicate bands of
interstellar medium dust and comet C/1995 O1 Hale-Bopp are included for reference. The
dotted vertical line indicates the position of the 9.8μm amorphous silicate band observed
in the interstellar medium. The red solid line indicates the best-fit models using a mixture
of amorphous and crystalline material with different grain sizes. The sharp 11.3μm feature
is evidence for crystallization, the shift toward longer wavelengths hints at grain growth
(Bouwman et al. 2001).
•
CO fundamental band at 4.7μm. The CO v=1-0 fundamental vibration-rotation
transitions at 4.7µm have been detected toward more than a dozen Herbig Ae and T
Tauri stars (e.g., Brittain et al. 2003, Najita et al. 2003). The lines often show a doublepeaked profile whose width varies from 5–10 km/s to more than 100 km/s and
correlates with the inclination angle (see Figure 3-7). Thus, a spectral resolving power
of R~50000 is needed to resolve these lines. The kinematical information provides
constraints on the location of the emitting gas in the disk. Both collisional excitation in
the inner dense warm gas (<1 AU) and resonance fluorescence in the outer disk (>5
AU) play a role in the CO excitation, and extended CO has been detected out to radial
distances of ~20 AU (0.3″ at 150 pc).
Page 23 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
•
Figure 3-7: CO v=1-0 fundamental vibration-rotation transitions at 4.7µm for five protostellar systems, observed with Keck/NIRSPEC at R=25,000. The green lines represent disk
model fits (Blake & Boogert 2004).
•
H3+ emission. The H3+ ion has strong emission bands at 3.9µm which have been
detected in the polar regions of Jupiter in our own solar system, and may be prominent
as well in exo-planetary atmospheres. A possible detection of H3+ lines in the HD
141569 transitional disk has been claimed by Brittain & Rettig (2002), but has not
been confirmed by subsequent searches (Goto et al. 2005). Nevertheless, like the H2
mid-IR lines, these are pioneering attempts to develop previously unexplored
spectroscopic diagnostics of the planet formation process, and MIDIR will push these
studies significantly deeper.
•
Organic molecules. Molecules such as CH4 (7.7µm), C2H2 (13.7μm) and HCN
(14.0μm) are key species in the organic chemistry that occurs in the inner (<20 AU)
planet-forming zones of disks. At the high temperatures in these regions, even more
complex, prebiotic molecules can be formed (Markwick & Charnley 2004), which
could be detected via their C-H stretching vibration in the 3–4 µm region. O-H and NH bonds also have signatures in this range. Note that ALMA cannot see symmetric
molecules without a dipole moment, such as CH4 and C2H2, which are prime building
blocks of organics. Models predict that the abundances of CH4 and C2H2 peak further
away from the star than that of the very stable CO molecule, so their widths are
expected to be narrower, requiring higher spectral resolution. The lines are usually in
emission, except in the case of edge-on disks. An excellent example is provided by
IRS-46 in Ophiuchus (Figure -3-8), for which the Spitzer-IRS spectrum shows strong
absorption by gaseous C2H2, HCN and CO2 in the 13–15µm region, indicating
temperatures of 300–900 K and very high abundances up to 10-5 with respect to H2,
orders of magnitude higher than in normal clouds. Its spatial extent is constrained to be
less than 20 AU and the width of the lines is Δv~10 km/s. MIDIR will be able to
spectrally resolve the mid-IR lines and to do spatially resolved absorption
spectroscopy against the disk photosphere.
Page 24 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure -3-8 Spitzer-IRS spectrum at R=600 toward the edge-on disk IRS46 in Ophiuchus
showing absorption from hot (>300 K), abundant organic molecules originating in the
inner disk (Lahuis et al. 2006). MIDIR will be able to both spectrally and spatially resolve
these lines in absorption or emission.
•
PAHs. Polycyclic aromatic hydrocarbons are the largest organic molecules that can be
observed in disks. Their strong emission features can completely dominate the MIR
spectrum. PAHs are not only interesting for the disk chemistry, but are also excellent
diagnostics of the UV radiation incident on the disk surface, and thus its flat or flaring
geometry (Acke & van den Ancker 2004). Moreover, they can heat the gas to
temperatures sufficiently high for photoevaporation, one of the primary mechanisms
for gas loss from disks (Kamp & Dullemond 2004). In contrast to thermal dust
emission, the PAH features are known to be extended to radii of at least 30 AU, as
demonstrated by spatially-resolved VLT spectroscopy (Habart et al. 2004, Geers et al.
2004). The spatial extent is expected to vary from feature to feature; for example, the
3.3µm feature is predicted to be more compact than the 11.3µm feature. These models
can be directly tested by spatially resolved spectroscopy with MIDIR.
•
Ices. At low temperatures (<90 K), the organic molecules will be frozen out as icy
mantles on the grain cores where they can also be studied through infrared
spectroscopy. Examples are solid H2O (3µm), HDO (4µm), CO (4.67µm), OCN–
(4.62µm), CH4 (7.67µm), NH3 (9.0µm) and CH3OH (3.5, 9.7µm) bands, which can be
observed from the ground. The line of sight through edge-on disks can also intercept
the cold outer layers of the disk where these molecules are frozen out. An example is
the edge-on disk CRBR2422.8-3423 in Oph whose spectrum shows very deep ice
absorptions (Thi et al. 2002, Pontoppidan et al. 2005). A spectral resolving power of at
least R~000 is needed to properly sample the line profiles, which vary from source to
source and contain interesting information on the ice environment and temperature
history. Moreover, high spectral resolution is essential to properly remove the telluric
lines, which limit the achievable signal-to-noise ratio. A S/N ≥ 100 is needed to put
limits on interesting ice ratios such as HDO/H2O for comparison with cometary and
solar-system data (Dartois et al. 2003, Parise et al. 2003).
Page 25 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
3.5 THE GALACTIC CENTER
The Galactic Center region is of great interest not only as the center of our galaxy but also
as the environment of the closest (quiescent) super-massive black hole. Largely
enshrouded by gas and dust, it can be best explored at radio, sub-millimeter, infrared, Xray and γ-ray wavelengths. All constituents of the inner few parsecs, the super-massive
black hole (SMBH), surrounding star clusters, streamers of ionized gas, molecular dust
ring and a supernova remnant have been studied extensively during the last years (See
Melia & Falcke 2001 for an overview).
Figure 3-9: The central parsec of the Galactic Center region observed at different
wavelengths. Left: VLT/NAOS-CONICA image at the NIR H (blue), K (green), and L (red)
bands (MPE/Clénet et al. 2004). Right: VLT/VISIR mid-IR image at 8.6μm (blue), 12.8μm
(green).
There are numerous topics of great scientific importance in this complex region, including
the stellar dynamics around the SMBH and its associated radio source Sgr A*, and the
formation and evolution of massive young clusters found in the immediate vicinity of the
SMBH. Figure 3-9 shows the near-IR and mid-IR view of the central parsec region,
illustrating their complementary nature. Both images are close to the diffraction limit of an
8m telescope at their wavelengths. However, MIDIR would provide approximately the
same resolution at mid-IR wavelengths than NAOS-CONICA in the near-IR, enabling the
first pan-spectral characterization of the region around Sgr A*.
One issue of particular interest is the accretion and emission mechanisms of the SMBH.
Variability and flares have been detected at infrared wavelengths (Genzel et al. 2003). The
intrinsic size of Sgr A* has remained unresolved at centimeter and longer wavelengths
because radio waves from Sgr A* are scattered by the turbulent interstellar plasma along
the line of sight. The scattering increases with wavelength as λ2, pushing observations to
shorter and shorter wavelengths. ALMA is expected to be a powerful tool for such
Page 26 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
observations, and MIR observations at the highest spatial resolution will enormously
contribute to this exciting puzzle. Recent broad band observations give upper limits of
factors of 2 – 10 above the theoretically expected SED for Sgr A*, based on a variety of
accretion- and jet models.
In the context of MIDIR it is important to note that the detectivity of Sgr A* is not limited
by the atmospheric background but by the diffuse emission of the surrounding gas and
dust. Thus, studies from space are severely limited by the low spatial resolution and the
high surface brightness. MIDIR on the E-ELT will be able to detect Sgr A* and measure
the broad band fluxes coming from the central ≤ 17 Schwarzschild radii at the most critical
wavelengths between L and N band (Figure 3-10). This will be essential information to
further improve models of black hole accretion processes.
Figure 3-10: Broad band spectrum of Sgr A* produced by a jet model, with a power-law
(PL) and a relativistic Maxwellian (MW) electron distribution, compared to radio and IR
observations (Melia & Falcke 2001).
In summary, the Galactic Center is an ideal target for MIDIR. It will be too bright for
JWST and can provide significantly higher angular resolution and much better time
resolution than MIRI.
3.6 THE LUMINOUS CENTERS OF NEARBY GALAXIES
The most extreme starbursts are found in mergers of gas-rich galaxies, where the
dissipative gas components quickly sink to the center of the potential well, resulting in an
intense burst of star formation. So-called ultra-luminous infrared galaxies (ULIRGs) are a
manifestation of this phenomenon, and approach quasar-like luminosities, which are,
however, almost entirely (re)radiated at mid- and far-infrared wavelengths. Although
locally rare, at high redshifts ULIRGs are responsible for a large fraction of the integrated
sub-millimeter background and the overall star formation budget.
Page 27 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
If ULIRGs are the progenitors of present-day elliptical galaxies the origin of the relation
between spheroid mass and nuclear black hole mass would be related to the ULIRG phase.
Many of the most luminous ULIRGs also contain an active galactic nucleus (AGN) but the
exact relation between the occurrence of an extreme starburst and the AGN activity is still
unclear.
The MIR wavelengths do not only penetrate dust but also provide numerous important
diagnostics: the ionic lines of [Ne II], [S IV], [Ar III], [S III] and others probe the photoionized gas, emission features of PAHs trace massive, young OB stars, the H2 lines probe
the warm molecular gas, and the broad 9.8 and 18μm silicate features contain information
on the absorbing material along the line of sight. Altogether, these diagnostics can be used
to separate the contributions from starburst and AGN to the total infrared luminosity. At
the wavelengths covered by MIDIR, the heating source of the dust is probed directly,
providing the link between the power source and the far-IR emission.
Observational studies of these important objects require imaging spectroscopy in the midIR at very high angular resolution and medium sensitivity. In the nearest ULIRGs, like
Arp 220 (Figure 3-11), the resolution provided by MIDIR will allow to spatially resolve
the individual components of AGN, supernova remnants, super star clusters and HII
regions, and other IR-luminous components. Most importantly, IFU spectroscopy will
permit the spectral classification and relative velocities of these components. It is
important to note that due to the extreme extinction most of these studies cannot be done
at NIR wavelengths.
Figure 3-11: A zoom in of the centre of Arp 220, the nearest ULIRG.
Figure 3-12: shows the velocity map of the centre of the Seyfert galaxy NGC 7582 in the
[Ne II]12.8μm line, taken with VLT/VISIR in high resolution mode. From the comparison
to models an upper limit on the black hole mass of 5×107Mo could be derived. To
appreciate the need for angular resolution we note that the nearest ULIRG, Arp220, is at a
distance of 75 Mpc. Even for a modest sample of ULIRGs one must already reach out to
distances of ~200 Mpc. At that distance, a 42m E-ELT will provide a maximum resolution
Page 28 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
of about 100 pc at N-band – still sufficient to reveal the large scale structures of the
nuclear regions, but not feasible from smaller space-based telescopes like JWST.
Figure 3-12: [Ne II] velocity map of the starburst/Seyfert galaxy NGC 7582, taken with
VLT-VISIR in HR mode (Wold et al. 2005).
The only other facility comparable to MIDIR in terms of resolution for this application
will be ALMA in its widest configuration at high frequency. However, ALMA will probe
the cooler bulk material at the Rayleigh-Jeans side, mostly neutral gas. MIDIR will
provide the important information at the Wien side of the Planck curve, mainly probing
the ionized gas, which directly responds to the illuminating UV radiation field, and
therefore contains essential information on the stellar population and/or the nascent AGN.
Hence, MIDIR will be the perfect complement to ALMA in this regard.
MIDIR will play a key role in uncovering the physical process relating starburst and AGN
in ULIRGs, and, by extension, the origin of the stellar spheroid–black hole mass relation
in galaxies. To be more specific, we list three example projects that would be enabled by
MIDIR:
a. The nuclear starburst region in ULIRGs is of the order of 0.3-1 kpc in size. These
regions will emit strongly in PAH emission. However, in the presence of an AGN, the
PAHs will be destroyed; thus their equivalent widths will be lower at the location of
the AGN. The spatial resolution provided by MIDIR is central to this application. This
measurement will even allow an estimate of the relative importance of starburst and
AGN for the overall energy output, with direct implications for the evolution of the
nucleus from pure starburst to AGN.
b. A clear tell-tale of an emerging AGN would be the detection of broad-line region
(BLR). Since the BLR is extremely compact, the high spatial resolution of MIDIR is
required in order to achieve enough contrast to spectrally identify a compact BLR on
top of more extended narrow line emission. This could be done using the Br-α
emission line at 4.05μm which shifts into the M-band for z ≥ 0.1, or using the Pf-α line
at 7.45μm, redshifted into the N-band for z ≥ 0.05, where the nearest ULIRGs are
found.
Page 29 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
c. High spatial resolution will enable the detection of nuclear high excitation lines which
would reveal a young AGN. The key diagnostic lines will be [Ar V] at 7.9μm and, in
particular, the [Ne VI] line at 7.6μm.
All of these will allow for the first time a direct separation of starburst and AGN in a
representative (i.e., fainter) sample of ULIRGs, and hence a clear quantitative probe of
their physical relation. This will allow tests of a whole range of models, from simple
intuitive questions such as the termination of starburst activity by the AGN feedback to
detailed physical models of radiation-pressure supported circum-nuclear disks. With its
combination of high spatial resolution, sensitivity, and access to the crucial mid-IR
wavelength windows, MIDIR will provide a unique probe of the relation between
starbursts and AGN, and likely provide more insights in the origin of the black hole –
spheroid mass relation.
3.7 AGN AT HIGH REDSHIFTS
Coronal lines are collisionally excited, forbidden transitions of ionic species with an
ionization potential of 100 – 400 eV. These lines can only form in extreme energetic
environments such as AGN, and are therefore considered good discriminants between
AGN and starburst dominated environments (Penston et al. 1984; Marconi et al. 1994;
Prieto & Viegas 2000; Rodríguez-Ardila et al. 2002; Reunanen et al. 2003). The strongest
coronal lines can be observed in the 1 – 40 μm range: [Si VI] … [Si IX], [Ne V], [Ne VI],
[Mg VII], [Ca VII], etc., and their strengths are comparable to that of lower ionization
atomic or molecular lines in this spectral range.
Coronal lines typically have asymmetric profiles and line widths of ≥ 1000 km/s, broader
than those measured in the narrow line region but narrower than those of the broad line
region. They are likely to originate from a region very close to the nucleus but still outside
the broad line region (e.g. Penston et al. 1984; Reunanen et al. 2003; Siebenmorgen et al.
2005: Yan et al. 2005), and remain mostly unresolved even at E-ELT resolution. Their
strength and exclusive ubiquity in AGN together with their low susceptibility to dust
extinction makes them ideal tracers of AGN activity, in particular in optical obscured
AGN at high redshift. Figure 3-13: illustrates the accessibility of coronal lines for a given
redshift.
At redshifts 3≤ z ≤6 and 2≤ z ≤4.5 the strong [Si VI]1.6μm and [Si VII]2.4μm lines,
respectively, fall into the N-band. The expected flux for these lines is about 10-14 – 10-15
erg cm-2s-1 at z = 0, which corresponds to a line peak of about 10-21 erg cm-2s-1Å-1 at
z~5. MIDIR will be able to detect these lines at the 3σ-detection within one night.
Page 30 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 3-13: Illustration of the detectability of the strongest coronal lines within the
atmospheric windows for a wide range in redshift. [Si VI]1.96μm and [Si VII]2.48μm are
particularly strong lines.
3.8 GAMMA-RAY BURSTS AT HIGH REDSHIFTS
Since their published discovery in 1973 gamma-ray bursts (GRBs) have been one of the
most exciting areas of extragalactic astronomy. Besides being extremely interesting events
in their own, they can be used to probe the ionization state and metal content of the
intergalactic medium (IGM) at high redshifts. For a short period of time they are the
brightest and most energetic events in the distant Universe, and are – even on E-ELT
scales – point sources. At z~10 an observed K-band wavelength corresponds to rest frame
UV light which may be strongly affected by extinction. All these points make MIDIR the
ideal instrument for GRB observations at high-z.
Figure 3-14 (left) shows the predicted flux densities from GRBs as a function of redshift.
GRBs observed after one day have 10μm flux densities of ~0.01mJy, which can be easily
detected. Within the first hours, GRBs out to z~15 have typical flux densities of 0.1mJy,
which is even within the capabilities of the MIDIR spectrograph. At that flux level (or
higher) one would expect a rate of 10-4 GRBs per square degree (Figure 3-14, right).
MIDIR spectroscopy may also be used to determine the redshift via the Pa-α (1.87μm) line
(3.0 ≤ z ≤ 6.2) or the Pa-β (1.28μm) line (4.8 ≤ z ≤ 9.5).
With its rather small field of view MIDIR is not well suited to discover GRBs – this task
has to be done by a dedicated survey satellite like SWIFT. However, once a high-z GRB
candidate has been identified, MIDIR would be able to observe the target much quicker
than JWST/MIRI.
Page 31 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 3-14: Left: Expected flux density of “typical” high-z GRBs at 0.5 (blue), 2 (red)
and 10μm (black) observed at 1 hour (solid), 1 day (dotted) and 10 days (dashed line)
after the burst. Right: GRB number counts as a function of flux density for different
observing wavelengths (both figures from Ciardi & Loeb, 2000, and personal
communication).
3.9 POLARIMETRY
Polarized light contains information on the emitting or scattering medium and thus
possibly crucial additional information on the target itself or its surroundings. Although
MIDIR, as discussed in this report, does not provide a sophisticated polarizing observing
mode such an additional capability is in principle possible (see discussion in section 4.6).
Here we discuss some science cases that require polarimetry.
Measurements of the geometry and degree of polarization yield important insights on the
physical processes and hence the conditions on the regions of interest. Specifically,
polarized light in the mid-IR may be due to:
•
•
•
•
Synchrotron radiation (linear polarization) for core-dominated radio sources produced
by Doppler-boosted emission from relativistic jets.
Cyclotron radiation (circular polarization) produced by electrons in a magnetic field.
Scattering (linear and circular polarization) – where the polarized state is dominated by
the last scatter – allows to study the geometrical and velocity relationship between
source, scattering medium and observer without spatially resolving the source. In
addition, the grain properties of the scattering medium can be determined from the
wavelength dependence of linear and/or circular polarization, which is of interest for
both resolved and unresolved geometries.
Magnetic fields (linear polarization) responsible for grain alignment that may lead to
polarized light in absorption (by the aligned grains) or in emission from the aligned
grains themselves. Circular and linear polarimetry of Zeeman-split lines can measure
the magnetic field strength and direction.
Page 32 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
3.9.1 Magnetic Fields in Star Formation
Magnetic fields play an important role in most processes responsible for star formation,
yet present knowledge of the magnetic field strengths and configurations is very limited.
However, knowledge of the magnetic field structure around the inner regions of young
stars is required to test the various predictions of magneto-hydrodynamic (MHD) models
of star formation. MIDIR would be able to detect the emission from hot dust in the
circumstellar disk, with the grains aligning with the local magnetic field. Studies to date,
mostly of high-mass stars, indicate a magnetic field that tends to be in the plane of the disk
and normal to the larger scale field in the flow (Aitken et al 1993). That the magnetic field
close to the star tends to be in the plane of the disk, is at odds with one of the more
plausible and attractive MHD models of bipolar flows (Pudritz & Norman 1986).
Unfortunately, the statistics are very limited and the spatial scale of the regions studied is
poorly defined. The high spatial resolution and extreme sensitivity provided by MIDIR
would allow determining the magnetic field geometries in unprecedented detail, e.g. the
degrees of twist and pinching of fields in the disk, drastically expanding the knowledge
base as well as the number of objects that can be studied.
3.9.2 The Structure of Young Stellar Disks
As discussed in Section 3.4.1 the direct imaging of circumstellar disks is of great interest.
Observations of the mid-IR polarized light strongly suppress the largely unpolarized light
from the central star, enhancing the contrast between the starlight and the polarized
scattered and/or emitted light from the protoplanetary disk. In this case, polarimetry will
improve the determination of the disk geometry and disentangle the various emission
components.
The structure of the circumstellar environment of Herbig Ae stars is a controversial
subject. Whether the disk is embedded in a roughly spherically distributed dust component
is debated and is object-dependent (Di Francesco et al. 1994). Unambiguously
discriminating between the disk and disk atmospheres versus envelope geometries requires
spatially resolved images and/or spectroscopy of the disk (Chiang et al. 2001). However,
adaptive optics systems still leave considerable flux in the wide wings around the central
core of a point source, which complicates the interpretation of surrounding, spatially
extended structures. However, the significance of the additional “AO speckle noise”
depends on the contrast between the central source and the surrounding structure. The flux
from the unpolarized central star (and its PSF wings) can be largely suppressed via
polarimetric techniques (Potter et al., 2000). It is possible to achieve at least a factor of 10
(and likely considerably more) reduction in the contribution from the central star without
the use of a coronagraph.
3.9.3 The Geometry of Active Galactic Nuclei (AGN)
In blazars and jets the main emission process is synchrotron radiation (sometimes Doppler
boosted) that is intrinsically polarized. In Seyfert galaxies the emission mechanisms are
more varied but scattering produces characteristic signatures that are observable only in
polarized light. Imaging- and spectro-polarimetry can be utilized to observe the central
regions of the AGN (e.g., Antonucci 1993, 2002), which is hidden from our direct view by
Page 33 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
an optically and geometrically thick dust torus. Nuclear continuum and broad emission
line photons can be scattered into the line of sight by electrons and/or dust grains located
along the polar axis of the torus, in particular in Seyfert-II galaxies. The polarization
behavior of AGN is believed to vary strongly with wavelength (Figure 3-15). This is
indeed seen in data reaching from the UV, optical to the near-IR. However, very little is
known about the polarization behavior of AGN in the mid-infrared, which would be the
most important wavelength range to study more heavily embedded AGN. Spectropolarimetry of the silicate feature at 10μm is a crucial diagnostic of polarization
mechanisms, but with existing telescopes this has only been possible for NGC 1068
(Aitken et al 1984, Lumsden et al. 1999). In NGC 1068, the polarization is due to warm,
aligned dust grains, with the position angle of polarization orthogonal to that at nearinfrared wavelengths.
Figure 3-15 Light scattered by material along the torus’s polar axis in type-2 AGN has an
observable polarization signature. Shown are dichroic absorption and emission by
aligned grains in a dusty disk for several inclinations: edge-on (solid line), 30° (dotted),
45° (short-dashed) and 60° (long-dashed) (Efstathiou et al. 1997).
Recently, Aitken et al. (2004) discussed a strategy for mid-IR polarimetry that facilitates
interpretation of such data. Their analysis points out that the polarization can arise from
emission and/or absorption from aligned grains, the polarization of which vary rather
differently with wavelength. Through this type of methodology one can discriminate
between thermal and non-thermal emission in AGN, disentangling the surrounding dust
from the AGN. (Spectro-)Polarimetry with MIDIR would revolutionize this area by
providing essential data on a large sample of objects.
Page 34 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
3.10 SUMMARY
An E-ELT aperture of about 42m would provide a spatial resolution 6.5× higher than what
will be achieved with JWST/MIRI at competitive point-source sensitivities. MIDIR’s
capabilities improve significantly on basically all areas of “traditional” MIR science, but
also open up MIR astronomy to areas which have traditionally been in the optical/NIR
regime. From the science cases we derive the following requirements:
•
Due to the huge thermal background, the large gain in sensitivity provided by an EELT will only be achieved for unresolved sources. Hence, the science case is strongly
biased toward compact sources and small-scale structures (small in angular units).
•
In most cases the size of mid-IR targets is in the order of arcseconds rather than
arcminutes. Since the distribution of MIR sources across the sky (within the
isoplanatic angle) is sparse a multi-object capability is not required. The study of
Galactic star forming regions requires the largest possible field for imaging, given by
the iso-planatic angle.
•
High-resolution spectroscopy (R~50,000 or 6 km/s) is strongly desired for several
reasons. First, to separate close spectral features (R>30,000 is needed to discern disk
emission from ambient nebular lines). Second, at R>50,000 the Earths orbital velocity
can Doppler-shift lines in and out of opaque, narrow windows. Third, higher spectral
resolution means better sensitivity to unresolved emission and absorption lines.
Besides, R≥50,000 will not be offered by any IR space observatory in the foreseeable
future.
•
One advantage of infrared over radio molecular spectroscopy is the coverage of
multiple rotational lines within a short infrared wavelength interval. The rotational
levels of the molecules in the thermal equilibrium are populated according to
Boltzmann's principle NJ/N ~ exp(-EJ/kT). The simultaneous measurement of multiple
lines therefore can serve as a practical measure of temperature, density, and the
fractional abundances of different molecules, including their isotopes.
•
For the spectrograph an integral field unit (IFU) is required for scientific and practical
reasons.
•
The wavelength coverage needs to include the thermal L and M bands, and the mid-IR
N and Q bands out to at least 20μm. The usefulness of the Q band beyond 20μm
depends strongly on the atmospheric characteristics of the telescope site. If the water
vapor content is low the spectral coverage should be extended to 27μm.
•
(Spectro-)Polarimetry would significantly add to the science case and add another area
to MIDIR not covered by JWST-MIRI. However, the trade-off study between added
scientific value and enhanced complexity is beyond the scope of this study.
In summary, MIDIR shall offer broad- and narrow-band imaging over a field
corresponding to the isoplanatic angle, low resolution long slit spectroscopy, medium
resolution (R~3000) IFU spectroscopy with wide instantaneous spectral coverage, and
high resolution (R~50000) spectroscopy.
Page 35 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4 General considerations
4.1 TOP LEVEL REQUIREMENTS FOR MIDIR
The Science Case in section 3 determines the following top level instrument requirements
which have been adopted throughout this document:
1. MIDIR shall cover the 3.5 – 20μm wavelength range in the atmospheric L, M, N, and
Q bands. It is a goal to extend the wavelength coverage to 27μm, depending on the
atmospheric transmission at the selected site.
2. MIDIR shall have four observing modes:
- A camera for broad and narrow-band imaging (~30 filters)
- A low resolution (R~300) long-slit spectrograph
- A medium resolution (R~3000) integral-field spectrograph
- A high-resolution (R~50000) integral-field Echelle spectrograph
[A possible fifth observing mode – (spectro-)polarimetry – is under investigation.]
3. The MIDIR instrument optics shall have diffraction limited performance (SR ≥ 0.8) at
all wavelengths and field positions.
4. The MIDIR AO system shall deliver at least 80% SR at N and Q bands, and at least
50% SR at L and M bands, with a goal to achieve even higher SR at LM bands. These
numbers are on-axis for a very bright guide star and an average V-band seeing of 0.8″.
5. In imaging mode MIDIR shall provide a field of view of at least the isoplanatic angle
at a given wavelength. In IFU spectroscopy mode the field of view shall be larger than
the seeing disk at a given wavelength.
6. The background from instrument + AO shall be well below the background from sky +
telescope at all wavelengths and spectral resolutions.
7. MIDIR shall provide parallel observing modes between:
- LM band imager and NQ band imager
- LM band spectrograph and NQ band spectrograph
- Imager and spectrograph at similar wavelengths.
4.2 CONSIDERATIONS ON DIFFRACTION LIMITED PERFORMANCE
The large aperture size of the E-ELT requires adaptive optics to correct for atmospheric
seeing even at thermal IR wavelengths. Figure 4-1: illustrates the importance of
atmospheric turbulence correction, which can yield an improvement of a factor of nine in
angular resolution!
Page 36 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
Seeing
HST
JWST
E-ELT
ELT-TRE-LEI-11200-0001
1.0
VISIR
max. resolution [arcsec]
1.00
0.10
0.01
0
5
10
15
20
25
wavelength [microns]
Figure 4-1: Seeing and diffraction limits for various aperture sizes as a function of
wavelength. Here we assume r0(0.5μm) = 12 cm (0.8″) and an E-ELT aperture of 42 m.
The figure also shows why mid-IR instruments on current 8m-class telescopes do not need
AO beyond about 17μm.
Diffraction limited performance in the case of MIDIR is defined as achieving a Strehl
Ratio (SR) of 80%. In order to evaluate the requirements for the diffraction limited
performance of MIDIR, it is necessary to investigate all elements which will prevent to
reach a SR of 80%. To first approximation, the Strehl Ratio at high Strehl is given by SR =
e-σ², with σ the wavefront error in radians, or SR = e-(2πδ/λ)², with δ the rms wavefront error
and λ the wavelength. The specifications and allowable errors for the various wavelengths
of MIDIR are given in the table Table 4-1.
Table 4-1: Specifications for AO system for the different bands. FOVs are given assuming
Nyquist sampling individually optimized for each band.
Wavelength Band
L
M
N
Q
Unit
Wavelength range
3.5-4.2
4.5-5.4
7.5-14
Reference wavelength for AO2
3.5
4.5
7.5
16
µm
Allowable RMS wavefront error
263
338
564
1203
nm
Diff. Limit (λ/D) for 42-m telescope
17.2
22.1
36.8
87.6
mas
Field Size3
34.7
45.3
37.7
40.2
arcsec
Maximum field angle4
24.5
32.0
26.7
28.4
arcsec
16-27 µm
2
The reference wavelength is taken to be the shortest wavelength of the band; If the AO system achieves its
performance at this wavelength, it will be achieved over the full band.
3
4k x 4k array mosaic (TIR) and 2k x 2k array mosaic detector, sampled at 2 samples/diff. limit (3.6 µm and
7.0 µm, respectively).
4
Radius of the circumscribing circle around the science field
Page 37 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
The most important elements to the wavefront error for MIDIR are summarized below:
Atmospheric Turbulence The standard atmospheric turbulence is caused by variations in
the index of refraction by temperature induced density variations in the atmosphere.
The turbulence spectrum is characterized by the typical cell size, r0, the outer scale, L0,
and the typical time scale of the variations in the turbulence, t0. A typical value for the
uncorrected wavefront for an E-ELT of 42 meter is of the order of 4 µm—not taking
into account the outer scale effects. The residual errors in the compensation of the
atmospheric turbulence will lead to deviations from the diffraction limit; see also
Fitting Error and Temporal Bandwidth.
Atmospheric composition The variation in the composition of the atmosphere introduces
again changes in the index of refraction. Especially in the wavelength range for MIDIR
the changes in the index of refraction due to variation in the H2O and CO2 content of
the atmosphere can potentially lead to large differential path lengths over the pupil of
the telescope and deviation from the diffraction limit. This element can give rise to
wavefront errors up to several microns, making it the most important error source in
MIDIR.
Index of refraction changes between sensing wavelength and science wavelength The
index of refraction is a function of the wavelength. This effect is most strong in the UV,
but also near strong absorption features the index of refraction can change strongly; the
absorption is the imaginary component of the index of refraction.
Residual Telescope Errors MIDIR will have to compensate for residual telescope errors
from various causes. The main mirror shape and co-phasing errors should be removed
by the active optics of the telescope, but residual errors at the level of several 10’s of
nm will have to be compensated for by the MIDIR AO system. Furthermore, all errors
due to vibrations of the telescope at frequencies above the cut-off frequency of the
active optics system need to be compensated by the AO system.
Non-common path errors Since MIDIR features multiple simultaneous beam paths, the
AO system can only partly correct for statical non-common path errors.
Furthermore, once an AO system has been implemented, the correction will not be perfect;
the wavefront errors above will in part be replaced by residual errors of the AO system, as
summarized below:
Anisoplantic angle and corrected field of view A given AO system will only correct
over a limited field of view, with a drop of performance outside this field of view. The
isoplanatic error is a major contributor for Single Conjugate AO (SCAO) systems. The
trade-off is between the isoplanatic error and the FoV surrounding the guide star. In the
case of a pure NGS SCAO system, this basically limits the targets to several arcseconds
from sufficiently bright NGS, i.e. the sky coverage is negligible. The anisoplanatic
error in a LGS SCAO reduces to the tilt-isoplanatic angle (and adds errors due to the
finite extend and height of the laser beacon), allowing for a significantly larger
separation between NGS and the science FoV and therefore much higher sky coverage.
For a Laser Tomographic AO (LTAO) system, the same advantages apply, but at a
significant increased FoV. The baseline system for MIDIR is a NGS SCAO system.
Since the field of the MIDIR imager is relatively large and the required correction high,
the anisoplanatic error is with about 500 nm rms wavefront error the largest error term
and drives the requirements of the AO system.
Page 38 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Temporal bandwidth of the AO system The temporal bandwidth of the AO system is
limited by the trade-off between the residual temporal errors and the limiting magnitude
of the guide star that is required. The residual errors by the limited temporal bandwidth
will lead to deviations from the diffraction limit. The residual error budget is divided
between the temporal, spatial and centroiding errors, leading to an error budget of
approximately 200 nm.
Spatial Sampling of the AO system The spatial sampling is again a trade-off, now
between the level of spatial correction of the wavefront—and therefore the fitting
errors—and the complexity of the system and the limiting magnitude which is
connected to the sub-aperture size.
Centroiding Errors The centroiding error is the error made in the determination of the
wavefront distortion and is mainly given by the brightness of the guide stars, either
NGS or LGS, structure of the guiding object and sensitivity of the wavefront sensor.
The full distribution of these error sources over the error budget is expanded in Section 5.2
4.3 THERMAL BACKGROUND
4.3.1 Why Chopping?
4.3.1.1 Background Noise Limited Instrumentation
Ground-based astronomical telescopes have to operate at or close to noise limits set by the
background radiation level imposed by the local atmospheric conditions. Cooling of the
mirrors is impossible as this would entail condensation and mirror seeing. Neglecting
Antarctica, potential ELT sites will have Tamb ~ 250 – 290 K. The optical surfaces of such
telescopes and the atmosphere will radiate both continuum and line radiation in the optical
path of any scientific instrument. Spectral features and intensities are elaborated elsewhere
in this document in great detail (cf Chapter 5). With various technical measures, especially
suitable transfer pupils in the cryogenic part of an instrument, this background radiation
can be controlled and reduced, but not avoided. For the temperature range given above this
radiation becomes noticeable for wavelengths greater than 1.6-1.7μm. Observations need
to be corrected for this background, i.e. the background needs to be measured and
subtracted. As the background signal is subject to classical shot-noise this noise enters the
noise of the pixel signal in the normal way i.e. by error propagation. Depending on the
pixel read noise of the specific detector this radiative shot noise contribution becomes
dominant under typical conditions longward of λ~ 1.7-1.9μm. This regime is normally
referred to as BLIP (Background Noise Limited Performance). MIDIR, operating at even
longer wavelengths is in all modes, even for highest resolution spectroscopy background
noise limited. To give an example, at λ ~ 10μm a pixel matched to Nyquist sampling of
the diffraction limit will be exposed to a background flux of ~1010ph/s and at λ~ 20μm the
pixel will see a background radiation of ~1011ph/s. Under such conditions it is still
possible, by careful subtraction and noise filtering, to detect astronomical objects 5-6
orders of magnitude fainter than this background level (c.f. Figure 4-2).
Page 39 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 4-2: Sketch of mid-infrared chopping and nodding technique of observations
(figure courtesy of the VISIR-consortium).
Using a practical example the classic way to acquire observational data in the extreme
BLIP-regime is demonstrated in Figure 4-2. At first, by fast modulation of the optical path
(“chopping”), 2 images are being acquired whereby the centers of the 2 images are
separated by typically 10-20arcsec. These 2 images are averaged and subtracted to yield a
“chopped -image”. Due to the modulation of the optical path slight asymmetries are being
introduced which result in a spurious signal orders of magnitude less than the back-ground
radiation, but still typically an order of magnitude bigger than typical astronomical signals.
This spurious signal in the chopped image is usually referred to as the chopping offset. In
a second step the telescope is moved, here perpendicularly to the chopping direction and
the chopping is repeated. Subtraction of the 2 chopped images cancels the chopping offset.
From this chopped and nodded final image, the 2 negative and the 2 positive images can
be extracted, re-registered and co-added to yield the final image. Note, that in this case the
useful size of the array is substantially reduced.
Page 40 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.3.1.2 General Idea of Noise Filtering
As described above, MIDIR can only detect astronomical objects with a meaningful
performance, if background signals are subtracted and if the associated noise is reduced to
the absolute theoretically possible minimum. The situation is further aggravated by the
fact, that the radiative background signal is highly variable with time. In the context of the
VLT design the frequency spectrum of these variations was quantitatively assessed by
Käufl et al 1991 [RD 2]. In the meantime, using TIMMI2 in burst mode these
measurements have been improved (c.f. Figure 4-3). Basically these measurements
confirm the general experience that chopping with typically 1Hz is sufficient to achieve
BLIP. The chopping offset (c.f. Figure 4-2) is very stable (most likely because it is entirely
due to radiation from the telescope) and nodding can be done without loss of performance
with time scales of 10-15 minutes.
Figure 4-3: This is a 2-dimensional representation of a typical measurement of the noisepower spectral density in the N-band as function of the wavelength. The data have been
taken with TIMMI2 at ESO's 3.6m telescope in grism spectroscopy using the burstmode
(figure courtesy M. Sterzik). This figure shows data at airmass 1.
Figure 4-3 clearly shows that the noise power spectrum correlates with the atmospheric
absorption features in this band. This is particularly obvious in the region of the ozone
band around 9.5μm, but also towards 13μm, where again substantial opacity exists. It can
be seen, that in “clean” parts, the power spectral density becomes is negligible below 1Hz,
which is in good agreement with the practical findings, that chopping with typically 1Hz is
sufficient to achieve BLIP performance.
Another reason for chopping is to reject interfering periodic signals. The most damaging
problem in this context is the temperature fluctuations of the detectors as a result of the
Page 41 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
closed cycle cooler expansion cycle. Those, however, can be reduced with passive and
active methods to a small enough level to become marginal.
In first approximation sky fluctuations should result in a common mode signal, affecting
all pixels the same, i.e. they should produce a constant offset signal or at least a signal that
can be approximated by a low order polynomial.
In conclusion one can summarize:
•
All infrared detectors working in the BLIP regime, i.e. λ > 3-4 μm deliver better
results if there is chopping with ~ 1 Hz
•
Without chopping, i.e. nodding only, one is left with residuals which can partially
be subtracted by software, especially if good flat-fields were available.
4.3.1.3 Optical Considerations, Panoramic Detectors
With the advent of panoramic detectors, there was great hope that the chopping
requirements could be waived. It turned out, however, (c.f. Figure 4-4) that the overall
stability of such detectors is not yet sufficient. The nodding process on a normal telescope
can not be done arbitrarily fast. After a movement of the telescope, even if it is only 10-20
arcsec the guideprobe(s) need to be readjusted and all control loops need to lock on and
close for the new position. This overhead for the VLT is at best ~10s and an ELT – having
many much more complex control loops - for sure will not be faster. In order to get away
with nodding only, changes of the gain of the detector and the differential flat field for
straight imaging should not exceed 10-5 in ~ 100s (see Käufl et al, 1991, [RD 2]). With all
infrared detectors ever tested in the BLIP regime at ESO (58x62 InSb in IRAC1, 64x64
As:Si in TIMMI, 240x320 As:Si-BIB in TIMMI2 and the 1024x1024 InSb in
ISAAC/CONICA) it was found that slow chopping is beneficial. For spectroscopy,
however, when it takes ~10 seconds or more to fill the detector pixels to a level that the
shot-noise exceeds the read-noise one finds that nodding only is sufficient.
The residual signal for “nodding only” in all cases is more complex than a common mode
signal modified by some flat-field (see Figure 4-4). There is some hope, that next
generation detectors will be intrinsically more stable. The Raytheon Aquarius array will
have special on chip electronics to reduce 1/f noise and it will have a better heat-sink part
of the chip-design, so that one can expect the chip temperature to be truly stable at the
milli-Kelvin level. However, one should not assume that for these devices chopping or any
other kind of signal modulation would not be helpful.
Page 42 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 4-4: Left: This is a typical example of a “nodded-only” observation taken with the
64x64 Si:Ga photo-conductor array in the ESO-TIMMI instrument. Shown here are 2
integrations 3 minutes each on-source and off-source subtracted. Apart from a gradient
the “noise” in this image has a certain structure; generally one finds under these
conditions a “print-through” of the read-out multiplexer topology.
Right: This figure shows the 2-dimensional power spectrum of a 64x64 TIMMI image. A
necessary condition of Background-Noise-Limited Performance (BLIP) is that the spatial
noise power spectrum would be “white”. This is clearly not the case for this data set. The
horizontal line in the data is due to an off-even effect of this particular detector. The
diagonal structure is not easily understandable. In any case based on this power-spectrum
a special adapted filter in Fourier Space can be developed to clean the data with limited
penalty, but decent rejection of the artefacts.
4.3.2 Some Background Information
4.3.2.1 Optical Problems of Telescope M2 Chopping
Even if there were no mechanical problems with chopping of an ELT M2 one has still to
consider, that the tilting of the M2 introduces optical aberrations: the chopping coma. For
the magnitude of the effect at the VLT see Noethe, 2003; (Annex B). Unfortunately, the
effect of image degradation is linearly proportional to the chopping angle. This entails,
that either both beams are half or one beam is fully affected by this effect. Strictly
speaking, for high-spatial definition (and high Strehl ratio imaging) the second beam can
not be used, in other words, the classical M2 chopping approach does sacrifice 50% of the
observing time at an ELT.
Page 43 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 4-5: FOCAL plane chopping: a kinematic mirror (e.g. a “Maltese cross” type
wheel, is moved in and out of the beam periodically. Thus the instrument will observe part
of the time the sky at the desired object location (“beam A”), and part of the time a
reference position on the sky (“beam B”). The mirror movement in the focal plane has as
consequence, that the sky reference exposure is out of focus. This configuration has been
notorious for a rather strong “chopping offset”.
However, telescope M2 chopping is not the only way for fast beam switching between
object position and “empty” sky. The first method is, to put a kinematic mirror into any
transfer pupil of the optical path. Indeed any tip tilt correction in an adaptive optics system
could serve the purpose, provided the tilt-associated aberrations are understood and
tolerable.
Another way is “focal plane chopping” which was preferentially applied in 'historic' times.
This method, however, reduces the efficiency of observing right away by a factor of 2, as
the reference image for sure can not be used as in case of M2 chopping (c.f. Figure 4-5).
Focal plane chopping, however, could be resurrected, if the detector were moved in the
focal plane.
4.3.2.2 Fundamental Mathematical Problems of Reconstruction
Let xi be a vector describing the true flux distribution coming from the astronomical object
and yi a vector describing the flux actually measured with a one dimensional array. Then
after chopping and nodding parallel to the array axis the two quantities are basically
related by the following equation:
yi = xi-j + 2 xi + xi+j (1)
For simplification it is assumed, that the chopper throw is exactly a multiple of the
detector pixel pitch; j is the number of pixels corresponding to the chopper throw.
Equation (1) above gives the entirely wrong impression, that there could be an easy and
Page 44 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
straightforward mathematical method to reconstruct the source brightness distribution xi
from the measured chopped and nodded image yj.
This becomes obvious if one resorts to Fourier space. Here a displacement along the xaxis corresponds simply to a multiplication with a phase factor e i k Δ . Δ is the spatial
extent of the shift (assuming a symmetric shift of +/- Δ around zero for the 2 chopping
positions. The relation between the Fourier transform of the source distribution ~
x and the
~
measured chopped distribution y is given by:
~
y (ω ) = e ikΔ ~
x (ω ) − e − ikΔ ~
x (ω ) = 2 ∗ sin (kΔ) ∗ ~
x (ω )
(2)
From this equation it is obvious that due to the chopping process all information at spatial
frequencies where sin(kΔ) gets small is basically lost. In a similar sense, the inversion of
the problem, i.e. the numerical solution of equ. (2) in Fourier space implies a divisions by
zero in Fourier space. It is thus easily understandable, why the reconstruction of the true
source distribution xi from the measured distribution yi in the presence of noise is not
possible with simple algorithms such as a matrix inversion. This leads, due to the divisions
by zero, to an explosion of the noise (Käufl, 1995 [RD 5]). Even if highly sophisticated
algorithms (Bertero et al. 2000 [RD 6] or Lenzen et al. 2005 [RD7] ) are applied, which
control the propagation of noise, one is still stuck with the fact, that chopping destroys all
information on spatial scales corresponding to the chopper-throw (and its “harmonics”).
The minimum requirement to recover this information is to perform an observation with
another chopping configuration (e.g. rotating the chopper position angle by 90o) basically
implies doubling the exposure time. This is important to keep in mind, if one wants to
analyze the figures of merit of any observing and signal modulation scheme for thermal
infrared instrumentation at an ELT.
4.3.2.3 Algorithms for Noise Rejection
Generally, Figure 4-4 gives an example, the result of an astronomical observation obtained
under high background conditions without chopping is characterized by fixed pattern
noise. The general experience at various observatories with different instrumentation
shows that the cumulative effect of this noise results in a reduction of the S/N ratio by a
factor of typically 2.5 if one is not chopping, but nodding only as fast as the telescope
allows. This kind of spatially structured noise, however, lends itself to rather
straightforward treatment with Fourier-filtering or more sophisticated methods such a
wavelet filtering. The approach is obvious from the second frame showing the power
spectral density of a 2-D Fourier transformation.
Another potential to reduce excess noise in beam-switching only, i.e. extremely low
frequency chopping, could be to introduce a numerical emulation of the sine-wave
detection which is being used in Lock-in amplifiers. This idea is described in Käufl et al.
1991 [RD 2] and should be tested as soon as the next generation of high-flux infrared
detectors becomes available.
Page 45 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.4 REQUIREMENTS FOR THE IMAGING AND LOW RESOLUTION
SPECTROSCOPY MODE
4.4.1 Imaging scale
Diffraction limited imaging sensitivity for background limited application is proportional
to D4 where D is the telescope diameter. Thus, to be compatible to space missions at MIR
wavelengths, any MIR instrument at an ELT should work near the diffraction limit. A
large ground based telescope is best suitable for high spatial (and spectral) resolution,
survey applications at MIR wavelengths are best done using space missions, which due to
their smaller aperture, lower resolution and smaller Background load, can provide much
wider fields of view. In consequence, the MIDIR concept presented here is focusing on
diffraction limited spatial resolution.
In Table 4-2 the required pixel sizes are given for different telescope sizes and Nyquist
sampling at 7µm and 3.5µm, respectively.
Table 4-2: Nyquist sampling pixel scale for different ELT sizes using 1kx1k MIR detector
and 2kx2k TIR detectors.
30m
42m
60m
Pixelscale (Nyquist at 7µm)
24.06mas
17.19mas
12.03mas 35 µm
FOV (1kx1k)
49.2arcsec 35.2arcsec 24.6arcsec 71.7 mm
Pixelscale (Nyquist at 3.5µm) 12.03mas
FOV (2kx2k)
8.59mas
6.02mas
Linear(f10)
17.5 µm
49.2arcsec 35.2arcsec 24.6arcsec 71.7 mm
If a mosaic of 2x2 2k×2k detector arrays is used for the 3.5 to 5.5µm band (InSb or
HgCdTe) and the same array of 1kx1k detectors for the 7 to 27µm bands (As:Si), the
resulting field of view is the same for all wavelengths.
The required collimator focal length is given by the maximum acceptable pupil image
diameter, which we assume here to be 50mm, and the telescope f-ratio, in the following
assumed to be 10. This choice does not depend on the telescope diameter.
The required focal lengths of the individual camera systems depend on the pupil diameter
(50mm) and the used pixel sizes. It does not depend on the telescope diameter, as long as
diffraction limited Nyquist sampling is the goal. For three expected pixel sizes the
resulting camera focal lengths are given in Table 4-3. The resulting f-ratios are quite
moderate. Pixel sizes of 18µm or 25µm are acceptable for the TIR-region, the pixel size of
25µm for the MIR region 7-27µm are at the lower end of the acceptable region: Nyquist
sampling for the Q-band would require an f-ratio of about 2.5, hard to realize with a TMA
solution. However, for MIDIR Nyquist sampling at 20µm is not proposed, we prefer to
realize a constant FOV over all wavelengths, living with over-sampling factors up to 2.8
for 20µm compare to Nyquist sampling.
Page 46 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 4-3: Required camera f-ratio assuming 50mm pupil diameter.
Nyquist sampling at
Pixel size
3.5µm
7.0µm
18µm
10.28
-
25µm
14.28
-
30µm
-
8.57
The position near the pupil image within the collimated beam is used to place the
cryogenic Lyot stop, the filter wheel and the grism wheel.
4.4.2 Filter Selection
The requirements of interference filters for the MIR wavelength region has been studied
intensively by the VISIR Astronomical Filter Consortium (VAFC) see e.g.
http://www.irfilters.reading.ac.uk/library/presentations/hawaii/index.htm
Figure 4-6: MIR filter set proposed for VISIR by the VAFC (see webpage given above)
Page 47 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 4-7: High Q-band filter set proposed for VISIR by the VAFC (see webpage given
above)
The following filter list is based on these recommendations. We added some typical L and
M-filters here, as used for ISAAC e.g.
Table 4-4: List of proposed filters
Identifier
Central λ[µm]
FWHM[µm]
FWHM[%]
NB3.21
3,21
0,05
1,6
NB3.28(PAH)
3,28
0,05
1,6
L
3,78
0,58
15
NB3.80
3,8
0,06
1,6
NB4.07
4,07
0,07
1,7
M
4,66
0,1
2
N1
8,6
1,4
16
N2
10,7
1,4
13
N3
12
1,4
12
NNB1
8,3
0,6
7,2
NNB2
9
0,6
6,7
NNB3
9,7
0,6
6,2
NNB4
10,4
0,6
5,8
NNB5
11,1
0,6
5,4
Page 48 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Identifier
Central λ[µm]
FWHM[µm]
FWHM[%]
NNB6
11,8
0,6
5,1
NNB7
12,5
0,6
4,8
PAH1
8,59
0,42
4,9
Ar III
8,99
0,14
1,6
SIV_1
9,82
0,18
1,8
SIV
10,49
0,16
1,5
SIV_2
10,77
0,19
1,8
PAH2_1
10,67
0,4
3,7
PAH2
11,26
0,59
5,2
SiC
11,85
2,34
19,7
PAH2_2
11,88
0,37
3,1
NeII_1
12,27
0,18
1,5
NeII
12,8
0,21
1,6
NeII_2
13,03
0,22
1,6
Q1
17,65
0,83
4,7
Q2
18,72
0,88
4,7
Q3
19,5
0,4
2,1
The collimated beam will have a diameter of about 50mm.
Directly ruled KRS5-grisms up to 100mmx120mm can be produced by Zeiss (Jena),
optimized for the L/M, N and Q-band windows. The exact grism dimensions will depend
on the final choice of the camera focal length and the required spectral resolution. Here we
assume an f-ratio of 14.3 (3.5µm) or 8.57 (7µm), respectively, and a pupil diameter of
50mm (as given by the preliminary optical design). This results in a focal length of about
714mm and 428.5mm, respectively. Given a pixel scale of 25µm and 30µm, the resulting
grism design parameters are quite moderate, for details see chapter 6.3.4.
4.5 REQUIREMENTS FOR THE MEDIUM AND HIGH RESOLUTION
SPECTROMETER
In the following discussion of the requirements for the spectrometer, we will start with
general considerations followed by specific requirements on the medium and on the high
resolution spectrometer separately.
Page 49 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.5.1 General Considerations
4.5.1.1 Spectral and Spatial Resolution and Sampling
The technical impact of the choice to get optimal sensitivity by making the instrument
near diffraction limited with proper sampling is big. The wavelength dependence of the
physical Airy disk radius limits the spectral range over which optimal sampling can be
achieved. This applies both for the spatial as well as for the spectral sampling. Apart from
the sensitivity issue, the aspect of expensive detector pixels makes oversampling very cost
ineffective. The Nyquist criterion indicates that at least two pixels are needed to digitize
the spatial variability in the signal, resulting in two pixels per FWHM or equivalently 5
pixels over the Airy disk diameter.
4.5.1.2 Options for Spectral Dispersion
There are various ways to obtain spectral analyzing power in an instrument. In general,
they can be categorized in four main principles:
1. Filtering, requiring different or tuneable filters (Fabry-Perot) [FP]
2. Dispersive (gratings, prisms or the like) [DE]
3. Fourier transform [FT]
4. Colour sensitive sensors [CS]
Option 1) is used in the imager for broad and narrow band imaging. However, going to
higher spectroscopic resolutions option 1) and also option 3) drop out because the high
variable atmospheric background and impact of absorption lines and emission lines on the
recovered spectrum result in very bad S/N performance and a very poor spectrophotometry. For option 3), the region with the worst S/N-ratio determines the overall
performance of the instrument, for option 1) it will be very difficult to obtain a reasonable
spectral coverage under similar thermal background conditions.
Option 4) is not yet possible, such sensors do not yet exist and medium and high
resolution spectroscopy will be very difficult to achieve. Thus, a classical spectrometer
based on dispersive elements will be the baseline for the study.
In a spectrograph the following techniques can be used to disperse light:
1. Prisms
2. Grating
3. Grisms
4. VPH (new)
5. Immersion gratings (new)
For mid-IR the variety of suitable refractive materials is very limited. Obtaining high
optical quality with those materials remains very critical. There is a justifiable reluctance
to use refractive optics in the mid-IR (options 1, 3, 4 and 5). In addition, the large OPD in
the optical path required for high resolution spectroscopy is difficult to obtain for options
1 and 3.
Page 50 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Immersion gratings might help considerably in reducing instrument dimensions; however,
this is not yet an available technique in MID-IR. Developments in this technology should
be followed and possibilities should be studied. The design in Chapter 6 shows however,
that the size of the collimated beam is strongly determined by the FOV combined with the
F-ratio. Small beams are very difficult to obtain and this limits the usefulness of immersed
gratings. The applicability of VPH at mid-IR wavelengths will be checked in the future.
Currently, option 2 (gratings) is strongly favoured, having the additional advantage of
keeping the optical system purely all-reflective.
4.5.1.3 IFU versus Long-Slit
The spectrometer will be of the integral field type of instruments. The additional
complexity of such an instrument is offset by the experience that will be obtained by all
integral field type instruments that currently are in use or being constructed. The
additional costs in detector pixels are considered to be consistent with the gain in
observation time efficiency and the gain in value of the science observations. A few
arguments are listed below:
2D-science targets:
• Point sources are often surrounded by an equally interesting structure (e.g. YSO or
AGN). Full spatial sampling with a long slit requires an enormous overhead in time.
• Even from small targets, strong emission lines might emerge from close by regions not
identifiable in general images due to strong continuum sources
Observational issues like pointing errors/accuracy and de-rotation:
• Uncertainty of source location in a sometimes complex environment
• Reduction of slit losses due to not sufficient accurate long term stability in pointing
• No image de-rotator required with an IFU. Field rotation with an IFU even improves
detector response (bad pixels)
• Easier photometry with IFU compared to single slit spectrometer
In an early stage during the OWL study, the conclusion was reached that for OWL scale
observations, a long-slit spectrometer will be too limited for the required science goals.
Cost arguments for this scale of instruments on the OWL observation platform are not
considered consistent as the cost trade-off between observation-time versus investment
strongly pushes to be most observation efficient. The same arguments hold for MIDIR and
the ELT.
Given our preference for an IFU, there are two types of integral field units for this kind of
instrument:
• Image slicing
• Fibre IFU
The latter requires fibre technology that currently is not available, whereas other
instrument are being developed based on image slicing in the proper wave length domain.
Currently, the image slicer is considered to be the optimal solution, however, also here,
developments need to be followed and studied.
Page 51 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.5.1.4 Slit/Slice Dimensions
Efficiency and resolution optimization has severe restriction on the wavelength range that
can pass through the system. For the imager, two parallel channels are foreseen, driven by
detector technology, proper spatial sampling and optical design complexity. For “slit”
spectrometers, similar arguments apply with the additional aspect of cutting the FOV by
“slits” or “slices”. Too wide slits/slices result in:
•
•
•
Big asymmetry between the final sampling spatially along slice and across slice
Big difference in spectral resolution for point sources against extended sources
Large instrumental dimensions for similar spectral resolution
Too small slices will lead to significant dispersion and reduce the efficiency of the
instrument. Part of this light can be recovered by oversizing the optics downstream of the
IFU, however, here are clear limits. In the MIRI design, this has been carefully analysed
and for MIRI the full wavelength (5-28 µm) has been divided into 4 channels, each
channel covering a wavelength range between λmin and 1.5*λmin (Figure 4-8) and using an
oversizing close to a factor 1.5.
45%
x oversize= 2
x oversize= 1.75
x oversize= 1.5
x oversize= 1.25
x oversize= 1
40%
% loss at grating
35%
30%
25%
20%
15%
10%
5%
0%
0.6
0.7
1.5*λ
1.5*λmin
0.8
0.9
1.0
λmin
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.0
Slice width (Units of λ/D)
Figure 4-8: The efficiency loss as function of slice width for various degrees of oversizing.
The slice widths are specified in Table 4-5 and have been defined based on the
atmospheric windows see Figure 5-1. Considerations for the chosen dimensions were:
L+M band use a different detector than the N- and Q band. The N-band is too broad for
the factor 1.5, but the outer-limits of the N-band start to be rather poor in atmospheric
transmission. Therefore, an optimal performance range has been introduced and the
reference wavelength for the slice dimension (λslice) is optimised for this range. Similar
arguments apply to the other channels. In addition, a small spatial undersampling is
accepted for the short area between λmin and λslice.
Page 52 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 4-5: Slice width definition of the spectrometer.
Band
L+M
N
Q
λslice (µm)
3.7
9
18
λmax(optimal)/λslice
1.35
1.48 1.38
4.5.1.5 Spectral and Spatial Sampling
Baseline choice for spectral sampling is Nyquist-sampling, i.e. two pixels per resolution
element (λ/D). Translated, the image of the slice on the detector will be sampled by two
pixels. Detailed analysis ensuring proper sampling including the line shapes of optics and
detector are outside the scope of conceptual studies.
Also for spatial sampling, the Nyquist sampling criterion is used: two pixels per resolution
element. Reference wavelength for the sampling is λslice, implying a slight undersampling
for wavelengths between λmin and λslice.
4.5.1.6 Conceptual Lay-out
Drivers for the design of the spectrometer channels are:
1. Efficiency
2. Similarity between the different channels
3. Weight and size (instrument needs to be cryogenic)
Ad 1) For spectrometers, the efficiency is usually limited, mainly caused by the number of
optical components and the diffraction element efficiency. Therefore, the efficiency of the
instrument should remain a constant issue in the design of the spectrometer. The most
critical element, the grating deserves therefore critical attention. Using the grating in low
order (preferably first order), prevents leaking of intensity into other, not used, orders and
prevents the need for order separation.
Ad 2) For preventing instrument complexity and additional development cost, it is prudent
to base the design for each channel on similar principles. As long as there is no need for
deviation, the channels can be kept similar, saving in development effort and in
development risks.
Ad 3) The weight and size of an instrument is always import, however, for cryogenic
instruments, the cryogenic complexity increases rapidly with increasing size and mass of
the instrument.
The resulting conceptual lay-out of the spectrometer is shown in Figure 4-11. After the
common pre-optics with the imager, a switch sends the collimated beam into the
spectrometer. The first part of the spectrometer consists of a dichroic filter system in
which the waveband separation takes place. After the waveband separation, the full FOV
will be re-imaged as one long slit in a for each waveband optimised IFU. In the IFUs, the
beams are re-imaged again to produce a well defined image at their exits and thus create
Page 53 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
access to an intermediate pupil and image to reduce the stray light generated in the system
and offer the required output beam parameters independently of the internal IFU
requirements.
The exit of the IFUs is an image of the slit that serves as input for the collimator system.
The collimator produces a collimated beam with sufficient diameter to ensure the proper
spectral resolution. The dispersion will be provided by gratings in almost Littrow
mounting, where the angle of incidence on the grating equals the angle of diffraction. The
camera makes an image of the spectrum on the detector array providing the required pixel
scale for optimised sampling.
Coll..
Grating
Camera
DetectorArray
Coll.
Grating
Camera
DetectorArray
Coll.
Grating
Camera
Detector-Array
IFU - LM
Waveband
selection
IFU - N
IFU - Q
Virtual slits
MR + HR mode
Figure 4-9: General lay-out of the spectrometer.
The spectrometer pre-optics, the dichroic separation into channels and the IFUs are
identical in the HR and MR mode, the detector arrays as well. This is clearly illustrated in
Figure 4-10. All channels work in parallel mode.
Page 54 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Collimator
Dichroic Switchyard
IFU-LM
IFU-N
IFU-Q
MR/HR switch
MR/HR switch
MR/HR switch
X-Disp
MR-Spec
X-Disp
MR-Spec
HR-Spec
X-Disp
MR-Spec
HR-Spec
HR-Spec
MR/HR switch
MR/HR switch
MR/HR switch
FPM-LM
FPM-N
FPM-Q
Figure 4-10: Block diagram of the spectroscopic part of MIDIR. Note that the various
wavebands operate independent of each other and can measure simultaneously.
The similarity between high resolution and medium resolution spectroscopy arms saves in
design, development, testing, and instrument complexity and thus most likely improves
instrument reliability. Nevertheless, there are issues that will be fundamentally different
between medium and high resolution spectroscopy.
4.5.2 Requirements for the Medium Resolution Spectrometer
Table 4-6 shows the specific requirements for the medium resolution spectrometer. For the
medium resolution the value of 3000 is preferred.
Table 4-6: Medium resolution requirements
LM-channel
N-channel
Q-channel
FOV
>1”×1”
>1”×1”
>1”×1”
Spectral resolution
R = 3000
R = 3000
R = 3000
Spectral range single exposure
At most two
exposures for full Nband coverage
4.5.2.1 Trade offs between Field of View, Spectral Coverage and Number of Pixels
In the IFU the FOV is re-imaged to a long slit. The FOV requirements for the various
wavebands might be quite different. In addition, the degradation in spatial resolution for
longer wavelength implies that for a similar number of resolution elements, the FOV will
be larger. Scientifically, this is not always required and a proper trade-off must be made.
Table 4-7 shows the IFU parameters for the various channels. The rows indicated by
Page 55 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
“detector” specify the detector parameters. The rows indicated by “science” show the
input requirements from science or earlier trade-offs. The rows labelled by “slicing”
calculate and tune the slicing parameters. Fine-tuning the required FOV and the number of
detectors, result in an optimised use of detectors and a FOV close to the required values.
For L+M, the choice has been made to try to get sufficient dark sky background, requiring
a FOV of 1”x1”. However, this FOV drives the number of detectors up and also the field
that the optics should handle. It is currently not known whether this FOV is feasible in one
system, this is subject of further study including options to increase the FOV to beyond
1”x1”. A more relaxed FOV in L+M of 0.8”x0.8” requires 2 detectors in spatial direction
and seems feasible. There are various ways to change this number, but it requires added
instrument complexity or relaxing the sampling requirements. This issue will be taken up
by the succeeding study.
For the N- and Q-band channels, the FOV is automatically larger (due to the sampling)
and it is easier to obtain larger values, limiting on one hand the number of detectors and on
the other hand reducing the complexity of the fast cameras, that are optically more
difficult.
Table 4-7: IFU and FPM parameters for the three channels of the MIDIR spectrometer.
The red numbers are input parameters.
Telescope diameter
Now implemented for the bands in MIDIR
Channel
Detector
Spatial
Spec
Spatial
Spec
Spatial
2048
1024
1024
1024
1024
#Pixels free from Edges
-
50
-
50
-
50
#Pixels between slices
-
5
-
5
-
5
(μm)
18
18
30
30
30
30
Space between two arrays
(mm)
2,65
2,65
TBD
TBD
TBD
TBD
λslice
(μm)
3,7
Slicewidth (λslice/D)
(marcsec)
18,2
#IFU-pixels for FOV
#slices
9
44,2
2,0
2,0
2,0
2,0
0,8
0,778
1,33
1,32
1,77
1,93
3770
1797
44
30
Obtained FOV
Final number of slices/resolution elem.
0,80
20,0
20
1,0
2
2
2
2
1
0,76
1,33
1,28
1,77
1,86
43
30
30
20
22
9,73
6,67
6,67
3,33
3,33
(mm)
76,4
3000
76,4
61,4
3000
61,4
61,4
3000
30,7
1
1
1
λblaze
(μm)
3,7
9
18
λ-range over detectors
(μm)
Dimension FPA
Resolution
2,5259
λmin
Science
15
44
parameters Order
full Littrow
22
9,73
Camera F/#
Grating
20
15,0
2,0
2
(arcsec)
874
22,0
2,0
Acceptable #detectors
88,4
2,0
integral #slices/detector
#detectors for required FOV
18
2,0
#slices fitting on one detector
Result
42
Q
2048
#Pixels per resolution element
Required FOV (arcsec)
Slicing
(m)
N
Spec
Detector Array size
Detector Pixel size
Science
L+M
3,072
λmax
λmin
6,144
λmax
λmin
λmax
Optimal range
(μm)
3,5
5,5
8
13
17
25
Extended range
(μm)
3,0
5,7
7,5
14,0
16,0
27,5
Page 56 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
In the spectral direction, the number of pixels available contributes to the spectral range
that will be covered instantaneously. The rows labelled “Grating parameters full Littrow”
(Table 4-7) show the spectral range corresponding to the number of detectors in spectral
direction for the medium resolution channels. From the table it can be inferred that at most
two exposures are needed to obtain coverage of the full wavelength range. Especially for
the N-band, this was considered essential, as stitching spectra around 10 μm should be
prevented.
4.5.2.2 Dependence of the Spectral Resolution on Wavelength
Figure 4-11 shows the general dependence of the spectral resolving power on wavelength
for spectrometers where the sampling is matched to λslice. For wavelengths smaller than
λslice, the pixel scaling starts to limit the performance of the spectrometer. Being able to
reconstruct the spectra with only 1 pixel sampling allows for unresolved sources to obtain
a higher spectral resolution. For wavelengths larger than λslice, the grating is limiting the
resolution. The sampling degrades as well (in exactly the same way as it does spatially).
This type of behaviour will be present in all spectral channels, and the averaged resolution
per spectrometer is lower than 3000.
3500
Resolution
grating resolution
limited
increasing oversampling
on detector
pixel scaling
limited
3000
2500
2000
1500
1000
500
0
0.0
0.5
1.0
1.5
2.0
2.5
Wavelength [λslice ]
Figure 4-11: The Spectral Resolving Power intrinsic to gratings, where all parameters are
optimized for λslice.
The extent how much of the spectrum will be sampled by the detector depends on the
required resolution and the number of pixels. For MIDIR, the full channel width cannot be
covered in one exposure. There are two ways to switch the waveband selection over the
detector: 1) scanning and 2) different grating parameters. Using the grating to scan over a
wider wavelength range, the resolution will just follow the curve of Figure 4-11. Another
option is to switch gratings, where you can redefine the grating parameters. This enables
you to get more homogeneous spectral resolution over the wavelength width per channel,
but the sampling issues on the detector will remain the same.
Page 57 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.5.3 Requirements for the High Resolution Spectrometer
High spectral resolution for very narrow emission or absorption lines is important for three
reasons. First, to resolve and separate close spectral features (a resolution of R >30,000 is
needed to simply discern circum-stellar disk emission from ambient nebular lines without
ambiguity). Second, at R > 50,000 narrow atmospheric windows can provide higher
sensitivity than the band average. Third, higher spectral resolution means better line
sensitivity, even for absorption lines, since the signal-to-noise in the continuum decreases
as √R for background-limited performance. This is impressively illustrated in Figure 4-12.
Figure 4-12: Model spectra of C2H2 at 900K and HCN at 600K (assumed Doppler
broadening ~4 km/s) at a resolutions of R=2000 (left) and R=50000 (right). Figure
provided by F. Lahuis.
Table 4-8 shows the specific requirements for the high resolution spectrometer. It should
be noted, that a lower resolution can usually be obtained by rebinning the spectra. For the
Q-band, it was decided to reduce the spectral resolution as it was deemed that the
requirement for the highest resolution was not very strong, whereas the technical
complication increases (requiring a grating twice the size compared to the N-band).
However, this does not imply that it is impossible to go to higher spectral resolution.
Table 4-8: High resolution requirements
LM-channel
N-channel
Q-channel
FOV
>1”×1”
>1”×1”
>1”×1”
Spectral resolution
R= 50000
R= 50000
R= 25000
V= 6 km/s
V= 6 km/s
V= 12 km/s
V= 12000 km/s
V= 6000 km/s
V= 12000 km/s
Spectral range
Page 58 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.5.3.1 High Resolution Dispersion
The arguments for the band selection and the FOV are exactly similar for the medium as
the high resolution spectroscopic mode. The sole difference is the amount of diffraction
needed. High spectral resolution will be obtained by creating a large optical path
difference (OPD) in the beam. The resolution can be obtained by
Formula: R = λ/Δλ = OPD / λ
For λ=10μm for R=50000, this requires an OPD of approximately 500 mm. This can not
be obtained by first order gratings, but gratings in Echelle mode, are feasible. High order
solutions need to be checked. This automatically means that overlapping orders become a
serious issue that needs careful analysis. The order separation can be accomplished by:
1. filtering
2. cross dispersion
The filtering has been dropped from our discussion, fixed filters tend to make the
instrument inflexible and many expensive and technically difficult filters are needed.
Another option might be via tunable FP-filters, where the gap is adjustable. However, a
cascade of filters is required to select one single wavelength range. This option requires a
high risk technology development, which might turn out to be not realistic. So, for this
study only the cross dispersion is considered.
4.5.3.2 Spectral Coverage
Using R = 50,000 on 2000-1000 resolution elements (4096-2048 pixels) already available
in the MR spectroscopy mode requirements, provides an astronomical important velocity
resolution of 6 km/s over a sufficient wide range of velocities (12000-6000 km/s). So no
additional requirement will come from the high resolution mode. Nevertheless, for a more
complete coverage in wavelength many exposures may be needed.
4.6 CONSIDERATIONS FOR POLARIMETRY
The science case for mid-IR polarimetry in Chapter 3 has demonstrated the scientific value
of the information contained in polarized light, ranging from distant galaxies to planetary
atmospheres. At the time of writing, neither the TMT nor JWST have plans for mid-IR
polarimetry, which presents an opportunity to afford a unique capability with MIDIR on
the E-ELT.
4.6.1 Introduction:
The various astrophysical processes (intrinsic and secondary) lead to different types of
polarized signals (see Table 4-9 and Table 4-10).
As polarization is a vector quantity, increased spatial resolution can boost the observed
polarization, exemplified through observations of the active galaxy, Cygnus A. Ground
based observations of Cygnus A measured the peak degree of polarization at ~5%
(Packham et al., 1998), but diffraction limited imaging from the HST at 2μm measured the
peak polarization at ~30% (Tadhunter et al., 2000).
Page 59 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 4-9: Intrinsic polarizing mechanisms.
Polarizing Mechanism
Polarization
Electrons with low values of v/c moving Circularly polarized
in a magnetic field producing cyclotron
radiation
Relativistic
electrons
synchrotron radiation
producing Linearly polarized, with E perpendicular to
B
Emission from aligned grains
Linear polarization with E perpendicular to
B
Zeeman-split spectral lines
Circularly and linearly polarized
Table 4-10: Secondary polarizing mechanisms.
Polarizing Mechanism
Polarization
Scattering off electrons or dust grains
Linear
polarization
with
E
perpendicular to the scattering plane
Scattering of linearly polarized radiation off Circularly polarized
non-Rayleigh particles
Scattering of radiation (of any state) off Circularly polarized
aligned grains
Radiation passing through a medium of Linear polarization by dichroic
aligned dust grains
absorption with E parallel to B
Polarimetry at mid-IR wavelengths has both advantages and disadvantages over the more
common polarimetry at visible and near-IR wavelengths. On the positive side,
instrumental effects due to telescope and instrument surfaces that are refracting or
reflecting at oblique angles are much reduced as compared to shorter wavelengths because
the polarization effects due to Fresnel reflection and refraction on dielectric and metal
surfaces is much reduced (although the differential phase shifts (wave-plate action) are
worse in the IR). On the other hand, the materials required for polarimetry such as
retarding wave plates and polarizing beam splitters are less common than those used at
shorter wavelengths. Nevertheless, several mid-IR instruments have included or will
include polarimetric capabilities (MICHELLE on Gemini-N, TIMMI2 on the ESO 3.6-m,
TNTCAM at WIRO, MLOF and IRTF, and CanariCam for the 10-m GranTeCan).
Page 60 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.6.2 Persistent Speckles
A topic of concern are persistent speckles in adaptive-optics corrected images. Since
polarimetry requires that at least two images be recorded in different polarization states, it
is crucial that these two (or more) images are recorded within a time frame that is short
compared to the characteristic time scales of persistent speckles. Since adaptive optics is
rarely perfect, it is prudent to record the images within a time frame that is short compared
to changes in the atmospheric wavefront aberration and sky emission changes. At midinfrared wavelengths, these time scales are on the order of 100 ms. A polarizing beamsplitter very close to the focal plane can be used to measure two polarization states strictly
simultaneously and thereby drastically reducing the errors induced by temporal variations.
Such approaches have been successfully used at visible and near-infrared wavelengths and
have also recently been included in the mid-infrared CanariCam.
Of course, persistent speckles will add photon noise like any other background signal, but
any type of measurement will be subject to this (additional) noise. Indeed, polarimetry has
a distinct advantage in that the persistent speckles only add to the (random) noise, but not
to the signal. For an intensity measurement, the persistent speckles contribute a systematic
intensity signal as well as the associated increase in (random) photon noise.
4.6.3 Design Considerations
For the imager CanariCam may be regarded as a test-bed for the technical implementation
of high accuracy mid-IR polarimetry and be applicable to the E-ELT. CanariCam
polarimetry (Packham et al., 2005 [RD 14]) will offer, for the first time at mid-IR
wavelengths, dual-beam polarimetry on a 10m class telescope making use of a cold dualbeam analyzer and cold half wave retarders. This increases very significantly the accuracy
of polarization measurements as compared to existing single beam polarimeters, and
eliminates the often dominant effects of sky transparency/emission and speckles.
For the spectrograph, a dual beam spectro-polarimeter using a Wollaston prism mounted
at the pupil of the IFU may be considered. This sort of design could provide the spectropolarimetric function for the medium resolution spectrometer in MIDIR. However, such a
design has the following negative impacts:
•
•
The entrance slit plane would double in length.
The prism would introduce a transmission loss of at least 10 percent and it would
be difficult to move it out of the beam.
• The insertion (and retraction) of several half-wave rotators in the beam requires a
sophisticated mechanism.
In summary, the scientific importance of mid-IR (spectro-)polarimetry with MIDIR is
recognized. The current MIDIR design does not rule out the implementation of an
optimized polarimeter. However, such an addition has a significant impact on the
complexity of the instrument. A detailed trade-off study of the pros and cons of
polarimetry and possible technical implementations go beyond the scope of the Small
Study and can be investigated in a follow-up study.
Page 61 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.7 DATA RATES
The Aquarius 1Kx1K Si:As array will have 64 parallel video outputs and can be read at a
frame rate of 150 Hz. The pixel rate per channel is 2.5 MHz.
The video signal will be amplified close to the focal plane by symmetric cryogenic
amplifiers which operate at a temperature of 70 K (Figure 4-13). The necessary bandwidth
of the preamplifier is provided by the OPA 356 from Burr-Brown (Texas Instrument). It
has already been tested with an infrared AO sensor (Figure 4-14). This amplifier provides
a bandwidth of 38 MHz and a noise of 5nV/SQRT(Hz).
Figure 4-13: Cryogenic preamplifier design.
The ADC board presently used in the NGC controller developed at ESO has 32 ADC’s
which can perform 16 bit conversions at a rate of 1MHz. Pin compatible 3 MHz ADC’s
are available but have not yet been tested. The 32 channels of 1 ADC board generate a
data rate of 1.3 GBit/s. Since each ADC board has a fiber link which has a bandwidth of
2.5 GBit/s, the data can be easily sent over the fiber link to the pci-bus interface of the
NGC linux pc which performs the preprocessing.
For two ADC boards serving the 64 channels of one Aquarius array only one fiber link is
needed. If a mosaic of 4 detectors is to be read out, 8 ADC boards and 4 2.5 GBit/s fiber
links are needed generating a data rate of 10 Gbit/s. Rael time co-adding and
preprocessing has to be performed in the linux number cruncher pc. Co-adding can also be
performed in the FPGA on the ADC board prior to transmitting data over the fiber link. 16
bit ADC conversion is not needed but convenient if available, since dc-offsets or dc drifts
can be easily dealt with if required.
Page 62 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 4-14: Comparison of different amplifiers at cryogenic temperatures of T=77K.
In Figure 4-15 the NGC controller with a basic board generating clock and bias voltages
plus an additional 4 ADC channels and a 32 channel ADC board are shown. In Figure
4-16 a basic board is shown in its backplane. Each backplane can accommodate 6 boards.
All components to read out a mosaic of 4 Aquarius arrays at the full speed of 150 frames/s
are already existing today and all components apart from the 3MHz 26 bit ADC have been
tested. The processing power required depends on the algorithms to be performed in real
time.
Figure 4-15: Basic board and 32 channel ADC board of NGC controller.
Page 63 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 4-16: Basic board of NGC controller with backplane.
For the maximum data rate Σ to be store on disk we provide upper limits for three cases:
(i)
The parallel operation of LM band and N band medium resolution
spectrometer. The LM band data come from a 2×2 array of 2k×2k detectors
and will be stored at ≤1Hz; the N-band data come from a 2×2 array of 1k×1k
detectors and will be stored at ≤10Hz. For the spectrograph we assume a 32-bit
ADC:
Σ = 2×2×(20482×1 + 10242×10)×32 = 1.8 Gbits/s
(ii)
The parallel operation of the L band and N band imager read at maximum
speed of 150 Hz (N) and 32 Hz (L) (“high time resolution mode”) with a depth
of 16 bits/pixel: Σ = 2×2×(20482×32 + 10242×150)×16 = 18.7 Gbits/s
(iii)
The parallel observation of the N-band imager (10Hz) and the N-band high
resolution spectrometer (1Hz) in normal operation mode:
Σ = 2×2×10242×(10×16 + 1×32) = 0.8 Gbits/s
We conclude that the data rates will be demanding on the data system, especially in the
high time resolution mode, but well feasible for a computer system 5 – 10 years in the
future, and even well below the demands by other astronomical instruments.
4.8 CALIBRATION: REQUIREMENTS AND SOLUTIONS
4.8.1 Introduction
Calibrations from the point of view of science are needed for two main reasons:
•
Control and stabilization of the observations
Page 64 of 204
Conceptual Design Study of MIDIR
•
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Mapping the observed parameters to physical quantities
The second aspect strongly depends on the success of the first. The difference between the
two is mainly related to the absolute knowledge of the physical properties of the sources.
The description therefore focuses firstly on the control and stabilization of the observation.
The observed parameters are relatively simple, images or spectra (integral field) on the
detector. However, to be able to analyse the data, the positions need to be characterised
and the relative fluxes need to be assessed. In addition, the data is distorted by varying
factors which have to be handled to improve the signal to noise of the data:
•
sky refraction, extinction and emission
•
telescope emission and transmission
•
detector variations and non-linearities
•
instrument and telescope performance and stability
To handle all these influences different technical solutions are needed. They will be
addressed in the next sections.
The quality of the calibration is specified by the calibration requirements. Those
requirements need to include the wavelength coverage, the optical quality, the field
distortion and stability, the wavelength calibration and stability, and the flat field quality
and (spectro-)photometric accuracy. For optimal performance of the AO system, an
independent calibration is required. Furthermore, by calibration of the non-common path
errors for the different channels a further improvement of the image quality can be
achieved. The calibration requirements may vary between the different channels. Their
exact specification will be subject to a follow-up study.
4.8.2 Variability of the Sky
The variability of the sky can be separated in different technological classes that for
technical and historical reasons are treated separately. In how far this distinction should be
maintained should be subject for study. With increasing angular resolution, these effects
tend to move more and more together and new issues do appear. Currently we identified
the following manifestations of the atmospheric variability:
•
Achromatic differential refraction through turbulent layers: Adaptive optics
[section 5.2]
•
Atmospheric dispersion (chromatic correction for different air thicknesses along
the lines of sight through the atmosphere) [section 5.1]
•
Differential emission from warm turbulent gas: Thermal background [section 4.3]
•
Strongly wavelength dependent variable dispersion due to water content (IR
radiation is very sensitive to the water content of the atmosphere) [section 5.1]
These issues are discussed in more detail in the sections indicated between the square
brackets.
Page 65 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
4.8.3 Telescope Thermal Background
The timescale of the variations in thermal emission from the telescope are slow with
respect to the atmospheric variability. Therefore, any system capable of handling the sky
variability should be able to correct for the telescope as well. However, the strategy for
coping with the atmospheric thermal emission might increase the sensitivity to changes in
the telescope thermal contribution. The chopping/nodding principle serves as illustration
of this effect. Chopping handles the atmospheric variability and nodding corrects for the
different instrument + telescope contribution caused by (slightly) different optical paths.
4.8.4 Detector Variations and non-Linearity
The detector variations are strongly dependent on environmental conditions and strongly
depend on the quality of the detector and its mount to stabilize these conditions. Provided,
the environmental conditions are stable, the response can be characterised pixel-by-pixel.
To ensure high performance operation, special attention should be paid which
environmental parameters need to be recorded together with the data to be able to correct
for changes in sensitivity. The calibration procedure should be able to mimic the different
conditions for the detector to calibrate this response.
4.8.5 Variability in Telescope and Instrument
Telescope and instrument are not completely constant over time. Changes in parameters
like pointing orientation might impact observation. The adaptive optics will handle part of
the optical path instabilities within the telescope and pre-optics of the instrument. Most of
the instrumental instabilities can be handled or controlled by proper instrument design.
Nevertheless, with the growing scale of the instruments more adaptive systems are
needed5. Parts of the system that can not be kept sufficiently rigid need active control
including a metrology system for a preferred closed loop operation. Currently, this is
outside the scope of the short design study. A detailed systems analysis should be
undertaken to trade-off risks in design and operation to the estimated performance
improvement of the instrument before the traditional practice of stiff and rigid optical
benches can be relaxed.
4.8.6 Instrument Characterisation/Calibration
For the calibration, the following parameters need to be characterised in MIDIR:
5
•
Optical quality of the whole optical chain (PSF and field distortion)
•
Detector performance, linearity, pixel similarity and response (flat fielding)
•
Spectral response of optical system
•
Polarisation dependent response
Examples: KMOS, X-Shooter
Page 66 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
For the spectrometer only:
•
Wavelength calibration and instrumental profile
The combination of an internal calibration system together with additional (limited) onsky calibration procedure should guarantee the performance of MIDIR.
4.8.7 Calibration Hardware Components
For the calibration of MIDIR we foresee a calibration unit in the warm pre-optics with the
following hardware components:
•
For calibration of the Point Spread Function (PSF) and field distortion:
An infrared diffraction-limited point-source (PS) on a XY-table that can be
positioned over the entire FOV of the imager and spectrometer. It should be usable
in combination with the monochromator M (see below).
TBD parameters:
- intensity,
- colour temperature,
- long/short-term stability,
- XY positioning speed and accuracy,
- warm-up time, power dissipation.
•
For flat-fielding:
An extended uniform blackbody source (BB), also covering the whole FOV of the
imager and spectrometer. It is not necessary that this source can be used in
combination with the monochromator, but it should be usable with/without
polarizers (see below).
TBD parameters:
- colour temperature range,
- stability,
- warm-up time, power dissipation,
- possibilities to monitor flux level e.g. with bolometer and filters.
•
For calibration of spectral response:
A tuneable monochromator, usable in combination with PS for every XY-position.
Full wavelength coverage of the L, M, N and Q bands will require more than one
grating order. The monochromator should be usable with/without polarizers in the
beam.
TBD parameters:
- spectral resolution (typically R ~10000 @ 10 μm),
- adjustable bandwidth (typically Δλ ~ 0.1-1.0%),
Page 67 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
- accuracy of the wavelength calibration (typically ~ 10-4).
•
For calibration of wavelength scale and instrumental profile:
A selection of gas cells that can be inserted into the beam at such a position that
combination with PS, BB and monochromator is also possible. If needed
(depending on the characteristics of that gas spectra), one cell for each spectral
band L, M, N, Q should be included. Switching of the cells into and out of the
beam should be so fast that efficient λ--calibrations can also be made on sky.
TBD parameters:
- choice of gasses (NH3, CO2, …?),
- column length and gas pressure (order of few mbar to achieve line
widths with Δλ/λ ~ 10-5).
•
For flux calibration:
Relative flux calibrations can be made by combinations of PS, BB and
monochromator. Special tools for absolute flux calibration are not foreseen; this
should be done on-sky with standards.
•
For calibration of polarization-dependent optical throughput and detector
properties:
A set of polarizers that can be moved quickly into/out of the beam, with/without
the PS, monochromator and gas cells, and also on-sky.
TBD parameters:
- polarizing efficiency and throughput at the relevant wavelengths.
•
For the calibration of the AO system the same diffraction limited infrared point
source can be used, although the output wavelength range needs to be span also the
AO wavelength range.
Neither for calibration of the AO system itself (e.g. interaction matrix, offsets on
the wavefront sensor,...) nor for calibration of common path errors special
hardware is required, but a specific calibration mode is required--also in the
software modules for the science cameras--for calibration of the non-common path
errors.
4.8.8 Calibration Strategy
The final performance of the instrument is determined by a mixture of internal and
external calibrators. Table 4-11 provides an overview how several parameters can be
calibrated.
Page 68 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 4-11: Overview global calibration strategy
Lab
Instrument
Sky
Field
calibrations
Point source + XY-table
Known targets
Spectral
Gas cell & monochromator
Atmospheric lines
Flux
Detector
characterisation
Relative flux calibration with
Flat field & monochromator
Known spectrophotometric objects
AO
Initial calibration
AO
Final calibration AO +
Noncommon path
None
Page 69 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
5 Atmospheric Effects and Adaptive Optics
The sensitivity and spectral coverage of ground-based mid-infrared observations depend
strongly on the emission and transmission of the Earth’s atmosphere. The strong influence
of the atmosphere is illustrated in Figure 5-1 and Figure 5-2 and depends on the elevation
of the site. Table 5-1 lists the resulting atmospheric transmission bands. As discussed in
more detail in Annex B both, atmospheric transmission and emission improve
dramatically with increasing altitude, especially the wavelength regions between the
typical atmospheric window like 2.5-2.8µm, 5-8µm and beyond 30µm are significantly
profiting from high elevations. But even at the centre of the N-band, at 11µm there is a
gain of more than a factor 20 between elevation 2600m and 5100m. This situation is
fundamentally different from NIR and optical observations as long as they are not skybackground limited. Estimates of the background fluxes for both imaging and
spectroscopy are in Chapter 7.
Table 5-1: Wavelengths of the broad band atmospheric windows.
Band
L+M
N
Q
Wavelength range (µm) 3.5-5.5 8-13.3 17-25
Above 5000m and at low humidity even the strongest absorption features are no longer
saturated, so that the observed spectra can be calibrated for telluric absorption. In periods
of low atmospheric water vapour content, the range of 5-8 μm becomes accessible to
groundbased astronomy. Moreover, the effective atmospheric temperatures – e.g. for
Cerro Macon, one of the potential ELT sites – may be lower than suggested by the USstandard atmosphere. Some conclusions may be derived from recent spectroscopy with
CRIRES6. A detailed report – in collaboration with the meteorological institute of the
University of Munich – is in preparation, and preliminary results are in good agreement
with findings by other groups.
However, emission and transmission are only two threats to mid-IR astronomy with ELTs
from the ground. Other, potentially important factors include atmospheric dispersion,
atmospheric turbulence (seeing), and fluctuations in the water vapour content. In this
chapter we discuss the magnitude of these effects and derive an AO system suitable for
MIDIR.
6
CRIRES, (Cryogenic Infrared Echelle Spectrograph) is presently commissioned at ESO's VLT on
Paranal. Spectra focussed around 2.4–3.4μm have been taken in mid June 2006 at a spectral resolution λ/Δλ
= 106 to be compared to radiative transfer codes such as PcLnWin and Hitran using true atmospheric profiles
destilled from the European Center for Medium Range Weather Forecast database.
Page 70 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
midlatitude, R=250
Chajnantor
Paranal
Transmission
1,0
0,1
0,0
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Wavelength [µm]
Figure 5-1: Atmospheric transmission for two elevations, 2600m and 5100m, typical for
Paranal and Chajnantor, respectively.
Midlatitude R=250
0,0000001
Paranal
Chajnantor
Skybackground [W/cm2/arcsec2/µm]
0,00000001
1E-09
1E-10
1E-11
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Wavelength [µm]
Figure 5-2: Typical atmospheric emission for mid-latitude US-standard atmosphere at
2600m and 5100m, respectively.
Page 71 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
5.1 ATMOSPHERIC DISPERSION
For seeing limited astronomical observation, at optical wavelengths atmospheric
dispersion is a known effect producing significant image distortion especially at high
zenith angles. As distortion becomes smaller and the diffraction limit is growing with
increasing wavelengths, even for diffraction limited observation at 8m-class telescopes
significant deterioration of the Strehl ratio by atmospheric dispersion is expected only up
to near infrared wavelengths. However, the step from an 8m-class to a 42m-class telescope
reduces the FWHM of the PSF down to the mas region, thus, even at thermal IR
wavelengths this effect gives significant Strehl Ratio deterioration.
In Figure 5-3 the atmospheric dispersion is given for the 0.5-25 µm wavelength region as
recently published by R. J. Mathar (2004). These data are compared to the standard
formula for refraction index of air given by Seidelmann (1992). The Seidelmann formula
fits quite well for wavelength regions below 5µm, however, beyond this limit there is a
significant deviation, water vapour plays a significant role within the MIR regime, not
only concerning atmospheric transmission and emission, but also concerning image
quality.
R efractio n in d ex (n -1) o f Air
2.10E -04
2.09E -04
2.08E -04
2.07E -04
n-1
2.06E -04
2.05E -04
2.04E -04
2.03E -04
2.02E -04
2.01E -04
2.00E -04
0
5
10
15
20
25
W av elength [µm]
Figure 5-3: Atmospheric dispersion for dry air (red line) and moist air (blue line) (Mathar
2004). Steps between atmospheric windows have been neglected here.
In Figure 5-4 the resulting effect for broad band imaging is shown: While for dry air the
dispersion over an R=5 broad band filter is negligible for all wavelength beyond 5µm even
down to zenith distances of 60 degrees, for moist air this effect can reach or even exceed
Page 72 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
the diffraction limit of an ELT. In addition, there is a fast variation with wavelength, an
effect that can not be corrected for with standard ADC configurations.
It should be noticed that the dispersion given in Figure 5-4 are peak-to-peak values for the
extreme case of 60deg zenith distance, RMS-values are smaller at least by a factor of 3.
Atmospheric Dispersion (R=5) compared to the Airy Disk
for different ELT-diameters and zenith distances
10000
λ/D(30m)
Dispersion in mas (R=5)
1000
λ/D(42m)
λ/D(60m)
100
10deg
20deg
30deg
10
40deg
50deg
1
60deg
Mathar, 60deg, R=5
0,1
60deg, 3g/m3
0,01
0
5
10
15
20
25
30
Wavelength [µm]
Figure 5-4: The extreme case of atmospheric dispersion for broad band imaging assuming
misted air (orange line) (Mathar 2004) and a zenith distance of 60 deg. This is compared
to the diffraction limit (λ/D) of a 30m, 42m or 60m ELT. Steps between atmospheric
windows have been neglected here. In addition, the dispersion for dry air is indicated for
different zenith distances.
In summary, dispersion within the TIR and MIR bands is no longer negligible for the next
generation telescopes. However, these effects are near the diffraction limit of a 42mTelescope for extreme zenith distances and for relatively high atmospheric humidity.
Taking into account that anyway only the monotonous part of the dispersion can be
corrected by an optical ADC device, we recommend not including an ADC into the
science beam of MIDIR (the WFS-beam should be discussed separately). Instead, we
recommend keeping these problems in mind during the site selection phase, the ELT
equipped with a MIR facility should operate at highest elevation and lowest PWV content.
In addition, highest Strehl ratios will be reached for moderate zenith distances only.
The atmospheric dispersion should be addressed in more detail during the point design
study, the atmospheric dispersion within the MIR regime should be measured depending
on actual humidity profiles.
Page 73 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
5.2 ATMOSPHERIC TURBULENCE
The AO system for MIDIR is designed for median seeing conditions at a good site, and
allowing for off-zenith operation, i.e., a median seeing of 0.8”. For the design of MIDIR
AO system and performance estimates a scaled Paranal-like atmosphere was assumed with
the layer parameters specified in Table 5-2.
Table 5-2: Distribution of atmospheric turbulence.
Altitude (m) Fractional Cn2 per layer Layer wind speed (m/s)
0
0.335
12.1
600
0.223
8.6
1,200
0.112
18.6
2,500
0.09
12.4
5,000
0.08
8.0
9,000
0.052
33.7
11,500
0.045
23.2
12,800
0.034
22.2
14,500
0.019
8.0
18,500
0.011
10.0
In initial estimates the outer scale was taken to be infinite, with a value of L0 of 25-m for
full simulations. The atmospheric properties are summarized in Table 5-3
Table 5-3: Seeing properties for the various bands. Note that the K-band is included as
being the Wavefront Sensor band and that the outer scale is infinite.
Wavelength Band
K
Atmospheric model
L
M
N
Q
Unit
Scaled Cerro Paranal
Reference wavelength for AO
2.2
3.5
4.5
7.5
16
µm
Seeing (0.8" @ 500 nm)
0.60
0.54
0.52
0.47
0.40
"
r0 (0.16m @ 500 nm)
0.93
1.6
2.2
4.0
10
m
θ0 (2.3" @ 500 nm)
14
24
32
59
147
"
τ0 (3.4 ms @ 500 nm)
22
41
48
118
218
ms
Maximum TT angle
126
79
62
41
17
mas
Tip-Tilt Frequency
0.69
0.44
0.34
0.20
0.095
Hz
Tilt Isoplanatic Angle
56
89
115
191
408
"
Peak Wavefront Error
16
16
16
16
16
µm
Required DM Stroke
7.9
7.9
7.9
7.9
7.9
µm
Page 74 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
5.3 TIME DEPENDENT CHROMATIC EFFECTS
MIDIR will use a different sensing wavelength than the observing wavelength. Due to the
slightly different index of refraction at the different wavelengths additional errors are
introduced. In the case of MIDIR this means that the sensing wavelength is chosen as
close as technically practical, but some errors remain. According to Hardy7 the chromatic
errors can be split in four categories:
Angular dispersion Angular dispersion is the effect that incoming rays are—wavelength
dependent—bent by atmospheric refraction. This is not a typical AO effect and should
be compensated for by an atmospheric dispersion compensator (ADC), see also Section
5.1.
Chromatic path-length errors Chromatic path-length errors are errors induced by the
variation of the index of refraction as a function of wavelength. In correcting the
atmospheric turbulence, the same correction is applied for all wavelengths, leading to a
residual wavelength dependent error. The error is equal to σ ch2 = ε 2 ( λ , λ0 ) σ u2 , with σ u
5
the uncorrected wavefront, to first approximation equal to
the chromatic error coefficient give by
⎛ D ⎞3
1.03 ⎜ ⎟ and ε ( λ , λ0 )
⎝ r0 ⎠
λ0 n ( λ ) − n ( λ0 )
. The resulting rms wavefront
λ n ( λ0 ) − 1
error and Strehl as function of wavelength are plotted in Figure 5-4 & Figure 5-5
respectively. Since this error cannot be easily corrected, this counts as a (small)
additional error term, of the order of 100 nm over the full spectral range of MIDIR..
Figure 5-4: Chromatic phase error
produced by sensing the wave front at 2.2µm Figure 5-5: Chromatic Strehl Ratio
(black) or 0.589 µm (red), respectively, and produced by sensing the wave front at 2.2µm
(black) or 0.589 µm (red), respectively, and
observing at MIR wavelengths.
observing at MIR wavelengths.
7
“Adaptive Optics for Astronomical Telescopes, Oxford University Press, 1998, p. 322-325
Page 75 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Dispersion Displacement errors Dispersion Displacement errors are due to the fact that
each location on the telescope aperture receives rays that arrive from the same source,
but traverse different atmospheric paths. At finite spectral bandwidth the correlation
between path length errors at different wavelengths decreases, inducing additional
errors. This effect is at the level of < 10-4 waves over the full spectral range of MIDIR
for a sensing wavelength of 2.2 micron.
Multi-spectral errors Multi-spectral errors are caused by a lateral displacement of the
sensing and science wavelength. Again, Hardy [RD 9] has shown that for a sensor
wavelength > 1 micron this effect becomes negligible.
In conlusion, the total chromatic error term in MIDIR, when using a sensing wavelength
of 2.2 µm is small enough—of the order of 100 nm—to be taken as an error term and no
further complications of the instrument are expected. A shorter sensing wavelength would
lead to unacceptably, and potentially uncorrectable, wavefront errors and significant
performance degradation.
5.4 ATMOSPHERIC WATER VAPOUR
Changes in the composition of the atmosphere can play an important role, especially in the
Mid-IR, due to the large dependence of the index of refraction on the water and CO2
content. Currently the most accurate estimate of the water vapour composition is given by
Colavita et al. [RD 8], which gives an RMS variation in PWV of 11 µm for a 100-m
baseline. Assuming that the water replaces standard dry air and that the RMS PWV scales
5
⎛ D ⎞6
like Kolmogorov, i.e., the RMS scales as ⎜
⎟ • 11μ m , the RMS path length difference
⎝ 100 ⎠
can be calculated for each wavelength of MIDIR. The resulting wavefront error for the
MIDIR wavelength range is given in Figure 5-6. Assuming a correction of the wavefront
at either 2.2 or 10 µm, the resulting rms wavefront error and Strehl Ratio are plotted in
Figure 5-7 and Figure 5-8 respectively. As can be clearly seen, based on the above
assumptions, the water vapour turbulence contributes significantly to the resulting Strehl.
At the current baseline MIDIR design, using a sensing wavelength of 2.2 µm, the error
budget in the N band will be fully determined by the water vapour fluctuations, while the
performance in the Q-band cannot be achieved without additional wavefront correction in
the Q-band.
The contribution of the water vapour and CO2 fluctuations to the total turbulence is
severely hindered by a lack of information regarding the absolute value of compositional
fluctuations in the atmosphere for the different sites; the data above, from Colavita et al, is
based on indirect measurements using radar on Mauna Kea. Furthermore, no data is
available on the distribution of these fluctuations as a function of height and if these
fluctuations correlate with the temperature fluctuations or are independent.
This information can be largely obtained using both observations from interferometers—
for large scale fluctuations—as well as smaller stand-alone telescopes, which should be
part of a follow-up study.
Page 76 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 5-6: Residual wavefront error as a function of wavelength due to water vapour
concentration variation.
Figure 5-7 The residual wavefront error due to water vapour turbulence, when correcting
at 2.2 micron (black) and 10 micron (red).
Page 77 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 5-8 The resulting Strehl Ratio due to water vapour turbulence, when correcting at
2.2 micron (black) and 10 micron (red).
5.5 AO REQUIREMENTS AND PERFORMANCE
Development of a diffraction limited system down to the short end of the L-band would
require the development of a full XAO system, comparable to the AO system of the
Planetfinder concept. This falls outside the scope of MIDIR and would also put very
stringent requirements on the remainder of MIDIR. The requirements for the AO system
are given by the top-level requirements in section 4.1. Those requirements will permit
excellent imaging and contrast in this wavelength region,
Also, MIDIR will use a natural guide star for the Wavefront Sensor (WFS). In this
wavelength range efficient detectors exist, the difference in refractive index is minimized
and the WFS does not take away light from the science channels. For the wavefront sensor
we assume a standard IR detector with <5e- noise, >1kHz readout rate, sky background
mag. 13/sq" (~15,000 ph/m2/s). For these estimates a Shack-Hartmann WFS was assumed.
The limiting magnitude improvement of a Pyramid WFS needs to be investigated.
The error budget trade-off is shown in Figure 5-9. Given the maximum allowable error of
564 nm for a Strehl Ratio of 80% and fixed errors due to design choices, the remaining
errors follow.
Page 78 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 5-9: Error budget decision tree
Taking the above design parameters and assuming that there will be a certain margin to
allow for the finite outer scale, Miska Le Louarn performed a numerical simulation
including most effects to determine the resulting system performance. The resulting Strehl
Ratios for different wavelengths, on- as well as off-axis are tabulated in Table 5-4. The
resulting Point Spread Functions (PSFs) are pictured in Figure 5-10.
Table 5-4: Expected performance of the MIDIR AO system.
Wavelength On-axis SR 5” off-axis SR 10” off-axis SR
2.2 µm
0.36
0.33
0.26
3.5 µm
0.65
0.62
0.57
7.5 µm
0.89
0.88
0.87
10 µm
0.93
0.93
0.92
27 µm
0.99
0.99
0.98
Page 79 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 5-10 PSFs for MIDIR for median seeing conditions at zenith for a 42x42 actuator
AO system, running at 500 Hz update rate on a K-magnitude 10.4 compact object.
The simulations include the general error terms, assume an atmosphere with finite outer
scale and given parameters for the AO system, but do not include all chromatic errors,
errors due to variations in the composition of the atmosphere or non-common-path errors.
As can be seen from the table and figure, the requirement of a minimum Strehl Ratio of
0.8 are achieved down to 7.5 micron, and even at 3.5 micron a well defined diffraction
limited core is achieved over the full field.
Table 5-5: summary of the AO requirements.
Parameter
Linear number of actuators
Total number of actuators
Value Unit
42
#
~1400 #
Update rate
500
Hz
Limiting magnitude
~10
Kmag
WFS Wavelength
2.0
µm
Required DM Stroke
20
µm
Required TT Stroke
10
µm
The large stroke required is currently only achievable with DMs with large interactuator
spacings. This means that the DM for MIDIR needs to be either a deformable element in
the telescope or a system with two DMs; the first has a large stroke and low number of
actuators for reducing the required stroke and could also provide the Tip-Tilt correction,
Page 80 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
the second provides high-order correction, but at much smaller stroke. It is still to be
investigated how close both corrective elements have to be to the pupil.
Page 81 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6 Conceptual Design
This chapter provides the conceptual designs for the most critical system components: the
AO relay, pre-optics, imager, medium- and high-resolution spectrographs, cryostat and the
mechanical setup. It also discusses important considerations on calibration, detectors, and
possible chopping schemes.
6.1 ADAPTIVE OPTICS RELAY OPTICAL DESIGN
For this preliminary study, we studied the feasibility of the AO relay optics for two
telescope design options as mentioned in Section 8.2, namely:
•
an uncorrected F/4.5 intermediate focus (post primary and secondary mirrors) of
the 5-mirror architecture.
•
a fully corrected F/16 focus of a Ritchey-Chretien design.
6.1.1 AO relay preliminary specifications
Taking into account the specifications for the imager, we aim at a 40”x40” field.
Input F/#: 4.5 or 16, depending on the chosen telescope design. Those 2 options are
considered below.
Output F/#: 10
Image quality: as diffraction limited as possible in order to decrease the extra correction
possibly needed by the imager and spectrometer optics.
Deformable Mirror (DM) diameter: 20 cm diameter class
Exit pupil position further than 10 meters upstream from the final image plane.
6.1.2 Preliminary Optical Concept for a F/4.5-F/10 relay
An initial double-paraboloid design was initially considered and implemented for a 1:1
relay, meeting the specifications in terms of image quality. However, the F/# conversion
and the residual aberrations after the first 2 mirrors of the telescope made the adaptation of
such a concept highly difficult.
A different approach was chosen, consisting of a re-imaging of the pupil on a DM
mechanically located at the intermediate focus (with a central hole in the DM structure,
compatible with the central obscuration of M2 in the telescope architecture), followed by a
mirror setup allowing image correction and F/# conversion.
6.1.2.1 Layout
Figure 6-1 shows an example of implementation of such architecture:
Page 82 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-1: AO relay architecture
Table 6-1: Components table for the F/4.5-F/10 relay option.
Component
Type
Size
Comments
AO-M1
Even asphere
φ 271mm
-
DM
Deformable flat mirror
φ 208mm
Central hole
AO-M2
Off-axis even asphere
φ 242mm
-
AO-M3
Off-axis even asphere
φ 258mm
-
AO-M4
Off-axis toroidal
φ 240mm
No asphere component
6.1.2.2 Performance
Figure 6.2 shows the spot diagrams relative to the above-mentioned design. The circle
indicates the Airy diameter for a wavelength of 3.5 μm. The distortion is less than 1%.
The exit pupil is located about 25.3 meters before the relay image plane. A residual field
curvature of R=610mm at the AO image surface is compatible with both the imager and
the spectrometer design.
Page 83 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-2: AO relay spot diagrams for a point on the axis and at the 4 corners of a
40”x40” field. Airy disk is indicated for λ=3.5μm.
Figure 6-3 shows the wavefront error of the AO relay optics, significantly below the
diffraction limit.
Figure 6-3: RMS wavefront error of the AO relay optics
Page 84 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.1.3 Preliminary Optical Concept for a F/16-F/10 relay
As for the previous relay design, a double-paraboloid design was initially considered and
implemented for a 1:1 relay, meeting the specifications in terms of image quality. The F/#
conversion and the increased input field yielded by the F/16 telescope plate scale made the
adaptation of such a concept highly difficult.
The increased input field size also prevented a similar approach to the F/4.5-F/10 relay,
namely with a central hole in a DM mechanically located at the telescope focus plane.
A different approach was chosen for this draft design, derived from the double-paraboloid
architecture, replacing each parabola by a set of 2 mirrors.
6.1.3.1 Layout
The picture below shows an example of implementation of such architecture:
Figure 6-4: AO relay architecture
Table 6-2: Components table for the F/16-F/10 relay option.
Component
Type
Size
Comments
AO-M1
Toroidal
φ 324mm
No asphere component
AO-M2
Even asphere
φ 306mm
-
DM
Deformable flat mirror
φ 220mm
-
AO-M3
Off-axis even asphere
φ 360mm
-
AO-M4
Off-axis toroidal
φ 340mm
Conic
Page 85 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.1.3.2 Performance
Figure 6-5 shows the spot diagrams relative to the above-mentioned design. The circle
indicates the Airy diameter for a wavelength of 3.5 μm. The distortion is less than 1%.
The exit pupil is located about 31.5 meters before image plane. A residual field curvature
of R=1730mm at the AO image surface is compatible with both the imager and the
spectrometer design. Figure 6-6 shows the wavefront error of the AO relay optics,
significantly below the diffraction limit.
Figure 6-5: AO relay spot diagrams for a point on the axis and at the 4 corners of a
40”x40” field. Airy disk is indicated for λ=3.5µm.
Figure 6-6: RMS wavefront error of the AO relay optics.
Page 86 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.1.4 Considerations on the Impact of Chopping on the Optical Design
In order to implement chopping in MIDIR, the AO relay would have to allow a wider field
of view than presented above (without necessity for a diffraction-limited performance in
the part of the field aimed at the background). Depending on the technique that will be
chosen for chopping and the parameters thereof (namely, position and size of the
additional field required), the design of the AO relay can be adapted in terms of
component size and mechanical configuration.
As an example, let’s consider chopping by a full size translation of the detector along the
instrument focal plane:
- consequences for the f/4.5-f/10 AO relay structure:
The optical definition of the optical components does not change. However, the translation
of the detector needs to be performed along the direction perpendicular to the AO relay
plane in order to avoid vignetting. Furthermore, the dimensions of the optical components
need to be increased, as shown in the Table 6-3.
Table 6-3: Components table for the F/4.5-F/10 relay option with chopping.
Component
Size
AO-M1
φ 340mm
DM
φ 208mm
AO-M2
φ 300mm
AO-M3
290mm x 240mm
AO-M4
300mm x 220mm
- consequences for the f/16-f/10 AO relay structure:
As for the previous AO relay configuration, the optical definition of the components
remains the same. This time, no preferred direction is needed for the translation. The
component size is modified as shown in Table 6-4.
Table 6-4: Components table for the F/16-F/10 relay option with chopping.
Component
Size
AO-M1
440mm x 320mm
AO-M2
360mm x 280mm
DM
φ 220mm
AO-M3
340mm x 290mm
AO-M4
340mm x 280mm
Page 87 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.1.5 AO Relay Optical Design: Conclusion
The 2 draft designs described above show the feasibility of an AO relay system compliant
with the MIDIR requirements and the telescope interface at once. Presently, there are no
interface difficulties identified for the different telescope options (see section 8.2 of this
document for more information on those options).
6.2 OPTICAL: DESIGN PRE-OPTICS
This module comprises 2 distinct parts: a common path for the spectrometer and the
imager, and a specific pre-optics for the spectrometer which aims at shaping the optical
beam before the dichroic switch.
6.2.1 Pre-optics: Common path
6.2.1.1 Cryostat Window
Due to the large wavelength range that should enter the system, there is no single material
that provides optimum transmission with acceptable optical quality over the full
wavelength range. A typical material choice for the window is CdTe, with moderate
transmission extending from 0.3 microns to 25 microns. Its low thermal conductivity is as
well a valuable feature. Other materials as CsBr, CsI, Daimond, KBr, KRS5 are currently
investigated. Unfortunately, all materials have some disadvantages: They are too soft to be
properly polished, there refraction index is too high to achieve high AR-coating over the
whole wavelength range, they are hygroscopic or they are not available in larger
dimensions or appropriate homogeneity.
Due to these problems, an exchange mechanism could be a valuable solution: Using KBr,
or ZnSe for the 3.5 to 14µm range and switching to KRS5 or CdTe for the Q-band could
optimize the overall efficiency significantly. However, such a solution if realized to be
applicable within less than a minute, will be quite expensive. Ferro-fluidic sealing with
diameters up to 1m can be realized on request. Such a solution has been realized for TReCS (Gemini S) e.g.
The entrance window of the for-optics will be relative large. Ideas to use this entrance
window also as dichroic tighten the requirements for the window strongly because of the
variable surrounding pressure that will deform the window. A thin entrance window is
optically preferred. Ideas for foil kind, so extreme thin, entrance window look promising,
but requires investigations on the transmission, risk and other effects. Some calculations
are performed on 3 type of windows with a diameter of 100 mm (see Figure 6-7).
The analysis has been concentrated on the displacement due to varying atmospheric
pressure. Table 6-5 presents the max. displacement, the max. von mises stress and the
max. stress principle. The flat and spherical windows are out of cadmium telluride and a
foil made out of PET.
Page 88 of 204
Conceptual Design Study of MIDIR
Flat window
Doc. No
Issue
Spherical window
ELT-TRE-LEI-11200-0001
1.0
Foil window
Figure 6-7: The three different type of windows.
This short and brief analysis shows other than traditional windows are possible. Different
options could bring benefits in different situations and are worthwhile investigating.
Designing towards only internal push forces (negative max. stress principle) in stead of
pull forces (positive max.stress principle) is worthwile to study. Typical ceramic materials
withstand push forces much better then pulling. This might reduce wall thickness of the
entrance window. On the other hand foil windows will only withstand pure pull forces.
Foil windows are probably no option due to the risk of breakage, but a study of a curved
entrance window is interesting.
Table 6-5: Displacement of the three investigated windows under different atmospheric
pressures.
Max
Displacement
Max
Mises
mm
N/mm2
N/mm2
10mm @ 1000mbar
-7.6E-03
3.6
3.3
10mm @ 1020mbar
-7.8E-03
3.7
3.4
15mm @ 1000mbar
-2.6E-03
5.0
1.4
5mm @ 1000mbar
-39E-03
9.7
7.7
10mm @ 1000mbar
-7.1E-03
6.5
-6.5
-129E+03
3364
3245
Ø = 100 mm
Von Max
Stress
Principle
FLAT
SPHERICAL
FOIL
0,1mm @ 1000mbar
Page 89 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.2.1.2 Separation Imager/Spectrometer
Two alternative approaches can be followed, each one placing the separation between the
imager and spectrometer path either before or after the post-AO image plane.
A separation before the intermediary image plane (see Figure 6-8) presents the advantage
of an intermediary focus directly accessible by each path independently (thus requiring
only 1 field stop per path). The absence of any additional relay optics between the focus
and the imager and the spectrometer is also a positive point.
On the negative side, the space available for the collimator of the spectrometer is limited
and calls for careful 3-D mechanical design.
In the case of a separation after the intermediary image plane, the space constraints for the
spectrometer channel are eased, but to the expense of the introduction of an extra relay
optics (therefore additional optical surfaces in the system). This additional optics makes
the optical path significantly longer, which might add mechanical instability.
As a preliminary baseline, we choose to split the beam before the AO image plane (see
Figure 6-8), with a removable pickup mirror folding the central part of the field to the
spectrometer. Several options of parallel observing are possible in this configuration:
•
Dichroic split, observing in a certain wavelength range in the spectrometer and
imaging at other wavelengths
•
Use a small pick-off mirror selecting only the centre of the field, allowing the
imager to observe the field around a central obscuration. Note, that the border of
the obscuration will be gradual due to vignetting
Figure 6-8: MIDIR Pre-optics with separation before the intermediary image plane.
Page 90 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.2.1.3 Common Path Components Size:
We indicate in Table 6-6 the size of the optical components of the common path without
chopping taken into consideration, followed by an indication of the modified size if
chopping is considered (hypothesis taken: full detector translation in the instrument focal
plane).
Table 6-6: Optical components of the common path.
Component
Type
Size (w/o chopping)
Indicative size w/ Material
chopping
Cryostat window
Flat
φ 180mm
φ 320mm
CdTe
Pickup mirror
Flat mirror
φ 40mm
φ 80mm
Aluminum (e.g.)
6.2.2 Spectrometer Collimator
The limited field covered by the spectrometer as well as the F/10 beam delivered by the
AO relay allows the use of a single paraboloid mirror as first approach for the collimation
of the beam (cf Figure 6-9). If AO relay residual aberrations need to be corrected the shape
of the mirror can be chosen aspheric.
The collimator is followed by 2 dichroic plates separating the 3 spectral channels. The
maximal footprint diameter on the dichroics is about 20 mm, and their tilt angle w.r.t. the
optical axis is 20°.
Cold stop position
Figure 6-9: Spectrometer pre-optics layout .
Page 91 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
One will note the potential cold stop position after the folding mirror in this layout. An
alternative would be to put the cold stop on the folding mirror itself, which is equally
feasible.
However, the pupil imaging after the second mirror of the telescope might require
complex pupil apodization and filtering (complementing a correction using the high order
asphere parameters of the collimator surface and/or the design of the post-DM part of the
AO relay), pleading for a more accessible cold stop surface.
The image quality is diffraction limited over the field as shown in Figure 6-10.
Figure 6-10: Spot diagram of the spectrometer pre-optics collimator, followed by a
paraxial model for a F/6 camera – The Airy disk is indicated for λ=3.5 microns.
In terms of wavefront error, we achieve the performance shown in Figure 6-11.
Page 92 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-11: RMS wavefront error of the spectrometer pre-optics collimator, followed by
a paraxial model for a F/6 camera
The size of the collimator without and with chopping being considered is shown in Table
6-7:
Table 6-7: Collimator
Component
Type
Size (w/o chopping)
Indicative size w/ chopping
Material
Collimator
Parabola
φ 30mm
φ 60mm
Aluminum (e.g.)
6.2.3 The Spectrometer Pre-Optics
The system for spectrally filtering and spatially slicing the three spectral bands defined in
Table 6-8, ready for detection by the three spectrometer channels described above, poses
an optical problem which is similar to one which has been solved in the Mid-infrared
Instrument (MIRI) for the James Webb Space Telescope. We propose a solution for
MIDIR which takes advantage of this heritage, thereby mitigating a number of risks.
These include the manufacture of an Integral Field Unit which can slice a diffraction
limited field whilst maintaining high throughput and excellent image quality, and
confidence that a chain of dichroics can be procured which will divide the wavebands
efficiently.
Page 93 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 6-8 The MIDIR spectrometer pass bands and fields of view.
Band
L+M
N
Q
Operational range [μm]
3.0 - 5.7
7.5 - 14.0
16 - 25
Optimal performance [μm]
3.5 - 5.5
8.0 - 13.3
17 - 25
FOV (across slice direction ×
along slice direction) [arcsec]
0.8×0.8 1.33×1.30
1.77×1.90
Number of slices in IFU
44
30
20
Slice width [milliarcsec]
18.2
44.3
88.5
6.2.4 The Dichroic Chain
In order to separate the three spectral bands defined in Table 6-8, we propose to use two
dichroic filters, each of which is designed to transmit long wavelengths and reflect short
wavelengths. Their performance can be inferred from the observed transmission and
reflection spectra of the MIRI dichroics. The measured cryogenic performance of two of
these are plotted in Figure 6-12, which shows them to come close to meeting the MIDIR
requirements even before any optimisation.
Figure 6-12: Measured transmission/reflection curves at cryogenic temperature for the
MIRI 3a (left) and 1c (right) dichroics. The nominal MIDIR passbands are marked on the
diagram.
Figure 6-13 then shows a paraxial schematic of the pre-optics needed for the most
constraining L-band channel of MIDIR. The N and Q band channels will have a similar
appearance. The scale of the diagram has been enlarged by a factor of 3 in the y-dimension
to illustrate the paraxial optics more clearly. The two dichroics can be placed at any
location in the 130 mm long stretch of the beam between the rightmost pupil plane and the
focal plane of the dichroic section, with ample space available for folding and separating
the reflected beams for each of the three channels (see Section 6.2.2).
Page 94 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-13: A paraxial raytrace diagram of the MIDIR L Band spectrometer Integral
Field Unit input optics, running from the telescope focal plane to the IFU slicer mirror.
For the L-band the sampling is 0.0182 arc-seconds per slice, with a physical slice width of
1 mm assumed. The F/6 telescope beam therefore needs to be magnified by a factor of 45
to obtain the correct sampling at the image slicer. This magnification has to be achieved in
three stages: in the dichroic optics, intermediate optics, and IFU pre-optics.
The magnification in the dichroic optics is limited by the maximum size of field of view
that must be transmitted by the dichroic filters (2 x 2 arc-seconds) and the maximum
diameter of the filters of around 30 mm. The path length in the dichroic section is also
constrained. In Figure 6-13, the distance from the MIDIR entrance pupil to the dichroic
section focal plane is ~130 mm.
Following the dichroic optics focal plane the intermediate optics are used to magnify the
beam to F/60. Finally the IFU pre-optics further magnify the beam to F/270 and provide a
re-imaged pupil at the entrance to the integral field unit (IFU).
The wavefront error introduced per dichroic filter across the 15 mm footprint which is
used at the dichroics in the MIRI design is typically 14 nm for a single dichroic substrate.
This figure increases to 28 nm with the addition of the dichroic coating. We confidently
anticipate that the effect of increasing the footprint diameter to up to 40 mm will result in
only a small increase in this coating dominated wavefront error and so the magnification
constraints on magnification and path length in the dichroic chain may be relaxed in a
future design.
6.2.5 The Integral Field Unit (IFU)
The IFUs proposed for MIDIR are a direct development of the JWST/MIRI IFUs (shown
in Figure 6-14), which use all-reflective diamond finished aluminium optics to slice the
rectangular field of view of the sky into between 12 and 22 narrow slices, which are then
stacked for presentation to the entrance focal plane of the spectrometer.
Page 95 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-14 A disassembled MIRI Integral Field Unit. The slicer mirror (whose diamond
finished surface is 22 mm from top to bottom) can be seen to the right and the re-imaging
optics to the left. The output are two rows of 11 sliced images of the IFU’s field of view.
The layout of the optics from the slicer mirror to the spectrometer entrance focal plane for
MIDIR are shown for the most demanding case of the L band IFU in Figure 6-15 and
Figure 6-16.
In addition to the pupil stop in the input beam, the MIRI IFUs include individual pupil
stops for each of the exit beams whose purpose is to eliminate cross-talk between the
sliced images due to scattering at the slicer mirror. These exit pupils (along with all of the
optics between them and the detector) are oversized in MIRI in the spectral/dispersion
direction by a factor of up to 2.5 when compared to their nominal width as prescribed by
geometric ray tracing. This is done in order to recover light which would otherwise be lost
by diffraction at the slicer mirror. The desirability of including this design feature in
MIDIR would be the subject of a further study, but as a rough guide, the throughput would
be reduced to roughly 75 % of its maximum possible value by using geometrically sized
pupil stops and spectrometer optics.
Page 96 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-15 MIDIR L band IFU input optics - side view. The extreme rays for only five out
of the 44 slices are shown for clarity. Slice 1 is shown in pink, slice 44 in blue.
The slice width is close to the Airy width (λ/D) for a point source in both cases, and so we
are able to take advantage of the analysis and test program for MIRI, which has
demonstrated that the effects of diffraction and slicing do not significantly degrade the
transmission delivered by the IFU, whilst the quality of the reconstructed final image is
diffraction limited as long as the physical slice length is not too large.
Figure 6-16 IFU output optics - top view. The imager slicer mirror is 44 mm wide in this
view. Only 5 slices which include the extremities of the field of view are shown for clarity.
Page 97 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
This desire to limit the slice length is the source of the most significant difference between
the two applications. This is because MIDIR requires double the number of slices that
were used in MIRI, but we wish to keep the physical slice width as close to MIRIs 1 mm
as possible in order to be able to use the same manufacturing techniques. We propose to
accommodate this change by introducing a high degree of anamorphism (an anamorphic
factor of 5) in the optics which project the image of the sky onto the image slicer mirror.
In this way, the physical length of the individual slicer mirrors is kept below ~ 25 mm,
which in turn minimises the aberrations that will be introduced into the sliced image. This
anamorphism is then fully cancelled in the optics which re-image the slicer mirror onto the
spectrometer focal plane.
Some relaxation in the trade-off between slice length and anamorphism factor can be
achieved by increasing the complexity in the slicer mirror design, namely by moving away
from the identical spherical mirrors which are used in MIRI to tailor the mirror’s
performance as a function of position in the field of view.
6.3 OPTICAL: DESIGN IMAGER
The optics proposed for the imager is a pure reflective optics, composed by a TMAcollimator system (pupil size 50mm diameter) and two following TMA-Cameras, both
operating in parallel or alternatively: A dichroic mirror splitting between TIR and MIR
can be replaced by a solid mirror.
A TMA-solution has been proposed to provide optimum resolution all over the
wavelength bands and to provide a very compact overall design. The internal Strehl Ratio
of this system is better than 90% for the central 10arcsec FOV for all wavelengths, and
better 80% all over the entire FOV and wavelengths.
6.3.1 The Collimator
Actually, the input pupil position is assumed to be 15m in front of the focal plane, an
image plane curvature of 120mm is adopted (actual AO-design). However, the TMAdesign can easily adapt for different values, if the AO-output beam is near telecentricity.
Actually, the input pupil position is assumed to be 15m in front of the focal plane, an
image plane curvature of 120mm is adopted (actual AO-design). However, the TMAdesign can easily adapt for different values, if the AO-output beam is near telecentricity.
Page 98 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-17: Collimator TMA focal length 500mm, input f-ratio 10. This collimator is
capable to collimate the full FOV of a mosaic of 2x2 2kx2k arrays (MIR) (2x2 1kx1k
arrays for MIR, respectively) to a 50mm diameter pupil.
6.3.2 The TIR-Camera
To illuminate a 18µm pixel array at 3.5µm with Nyquist sampling, the camera f-ratio
should be N = 2*pixelsize/λ = 10.3 . Thus, assuming an input pupil diameter of 50mm, the
resulting focal length is 514mm.
Figure 6-18: 3.5µm to 5.5µm TMA camera with pupil diameter 50mm, assuming a pixel
size of 18µm.
Page 99 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-19: 3.5µm to 5.5µm TMA camera with pupil diameter 50mm, assuming a pixel
size of 25µm.
6.3.3 The MIR-Camera
To illuminate a 30µm pixel array at 7.0µm with Nyquist sampling, the camera f-ratio
should be N = 2*pixelsize/λ = 8.6 . Thus, assuming an input pupil diameter of 50mm, the
resulting focal length is 428mm.
Figure 6-20: 7µm to 27µm TMA camera with pupil diameter 50mm, assuming a pixel size
of 30µm.
Page 100 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.3.4 The Grisms
MIDIR is proposed to be equipped with a grating spectrograph providing spectral
resolutions power between 3000 and 50000. Lower resolution spectroscopy will be
realized by grism spectroscopy just entering one of several grisms into the collimated
beam of the imager. Grisms for several spectral resolutions and wavelength bands can be
proposed here. Thus, the list of grism given in the Table 6-9 is a very preliminary set of
possible grisms. In addition, a double-prism is proposed to provide low resolution
spectroscopic information over the full spectral range in one single shot.
Table 6-9: Preliminary list of proposed grisms an double prism to be inserted into the
collimated beam of the imager.
Wavelength
region [µm]
Material
Groves
per mm
Prism angle
2pixel resolution
power
2.8 – 5.2
KRS5
28
4.80 deg.
R = 1400
7.0 – 14.0
KRS5
10
4.46 deg.
R = 500
15 – 27
KRS5
5
4.80 deg.
R = 250
2.8 – 4.2
KRS5
50
6.56 deg.
R= 2250
4.4 – 5.6
KRS5
40
8.25 deg.
R = 2000
7 – 26
NaCl/CsBr
-
16.7/7.1
R = 86-1380
Figure 6-21: 3D representation of MIDIR imaging optics. A TMA is proposed for the
common collimator and for the re-imaging cameras in both channels.
Page 101 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 6-10: Short description of imaging components
Item
Description
Size (square)
Curvature radius Conic
Collimator M1
Aspheric off-axis
120mm
-564mm -5.0
Collimator M2
Aspheric off-axis
70mm
-366mm -2.6
Collimator M3
Aspheric off-axis
100mm
-402mm -1.6
TIR-Camera M1
Aspheric off-axis
100mm
-686mm -0.6
TIR-Camera M2
Aspheric off-axis
70mm
1087mm 0.0
TIR-Camera M3
Aspheric off-axis
120mm
-474mm 0.03
MIR-Camera M1 Aspheric off-axis
100mm
-684mm 0.4
MIR-Camera M2 Aspheric off-axis
70mm
-1624mm 0.0
MIR-Camera M3 Aspheric off-axis
120mm
-383mm 0.1
6.4 OPTICAL: DESIGN HIGH RESOLUTION SPECTROMETER
6.4.1 Introduction
In this study, we limit ourselves to the optical design of the N-band spectrometer. The Nband has been selected because the requirements for the N-band spectroscopy were
clearest from the start of the study where we expect that the complexity for all
spectroscopic channels is more or less equal although the complexity shift to different
aspects in the design:
• Moving from N- to Q-band complicates the design in the sense that the camera must be
faster, on the other hand the FOV is much smaller, which in turn simplifies the system.
Nevertheless, after working out the N-band we expect that it will be very difficult to get
the proper F-ratio in the system. Including 50% oversizing, the required F-ratio for the
Q-band will be approximately F/2.2. Possible solutions are: (1) reducing the oversizing
in the optics slightly, (2) allowing more pixels across the Airy disk, (3) increase the pixel
size from the current 30 µm to 40 or 50 µm. For Q-band, it is expected that the limitation
in overall instrument performance is dominated by the atmospheric transmission to a
larger extent compared to LM and N, giving a larger performance budget within the
instrument.
• Moving from N- to L+M-band goes in the opposite direction, the speed of the camera is
slower but the FOV increases. Here, the FOV starts to be critical, but multiplexing of
spectrometer systems could solve the issue. The tightening of the optical tolerances
becomes critical in the L+M-band due to shorter wavelength of the light. Staying within
the optical budgets requires different production technologies as for normal mid-IR.
Especially, the form accuracy and the surface finish of the optical components might be
challenging and it is recommended to demonstrate the production capabilities far before
finalizing the optical design.
Page 102 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.4.2 Global optical design
Based on technical readiness, the spectrometers are based on traditional reflection
gratings. In the design of such a system there is ample room to trade between several
design choices, e.g. the order of the Echelle gratings, the principle of order separation. The
basic design parameters have been fixed using a paraxial model. The results of this
analysis are summarized in Figure 6-22 to Figure 6-24. The spatial design parameters
follow from the IFU discussions, see earlier in the report.
In the green boxes, the input parameters are given. These parameters relate to the pixel
size of the detectors, the physical size of the array, wavelength range to be covered, the
required spectral resolution (R). λR must be chosen equal or larger than λslice to ensure that
the grating is the limiting factor for the resolution. The yellow box contains the twiddle
parameters for selecting the proper combination of grating tilt angle (“blaze angle”), and
spectral order. The “blaze angle” is the nominal angle for the grating, “central order blaze
wavelength” is the wavelength at exactly the blaze angle for the “central order”. The
“Lowest order” is the lowest order that is used in the calculation, taking care of the upper
end of the wavelength range. Full wavelength coverage will be obtained by tilting the
grating, the extremes indicated by “Min. scan angle” (tilting towards the normal) and the
parameters “Scan step” and “#steps”. The “Scan step” is chose such that for all tilts and
orders, the subsequent exposures do overlap. The extreme scan angles should cover the
whole spectral range and allow proper overlap between the different orders. Subsequently,
the parameters are selected so that the total wavelength range for each channel is covered
as is seen by the graph. Selecting higher grating orders results in more orders close
together requiring less scan angle for the Echelle, however, this requires more tuning in
the order separation.
The results (orange box) are the required grating parameters, the collimated beam size and
the order to be used. The graphs in the spreadsheet are indicative, as many aspects are not
included, like off-Littrow angles, impact of oversizing and diffraction on resolution and
the like. While the latter will lead to increased resolution of the gratings, the impact of the
detector sampling remains the same.
As can be inferred from Figure 6-22, the N-band channel Echelle will be designed
working in orders 8 to 13, 35 exposures are needed for a full coverage of the N-band in the
high resolution mode. The spectral resolution varies between R=40000 and R=60000. At
wavelengths larger than 12.5 µm, the resolution drops below 40000.
There are two remarks for the L+M-band and Q-band HR mode As can be seen in Figure
6-23, the order for the LM Echelle is kept low, to prevent the need for the cross dispersion
as it is expected that high quality filters are not yet so difficult to obtain in this wavelength
domain. However, this requires some more careful consideration as the grating size
becomes very large. The spectral resolution for the Q-band (Figure 6-24) is reduced
(nominal R=25000) compared to LM- and N-band to keep reasonable sizes of the grating.
However, science should indicate whether this trade-off is acceptable. Bigger size gratings
will complicate the design, but if needed several options can be studied.
Page 103 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
N-BAND HR-MODE SPECTROSCOPY
Parameters:
OD Adjustable Parameters:
Detector:
Grating
Pixelsize
ps
30 μm
#pixels
70000
Blaze angle
1024
θ
60
#cal pixels
8
Central order
#spectral detectors
2
Central order blaze wavelength
Distance to next array
Db
?
μm
Lowest order
De
?
μm
Min. scan angle
8 μm
λopt,max
13,3 μm
Scan step
60000
50000
40000
30000
-5,7
o
1,7
o
7
8
9
10
11
12
13
14
15
λ (μm)
7
25
7,5 μm
7,5
λmax
14 μm
13,8
λslice
9 μm
R
λR
50000
10 μm
Grating parameters:
a
60,74 μm
L
288,7 mm
Dcoll
144,3 mm
20
Scan angle (degrees)
λmin
Central (slice wavelangth)
For λ
8
δα
Nstep
#steps
Full range
Resolution
10,52 μm
α
Optimal wave-range
λopt,min
o
10
λc
Performance:
Resolution
MIDIR
ELT-TRE-LEI-11200-0001
1.0
15
10
5
0
-5 7
8
9
10
11
12
13
14
-10
Cross dispersor:
λ (μm)
acrossdisp
19,33 μm
o
Scan range ( )
-5,7
4,5
Figure 6-22: HR grating parameters for the N-band channel.
MIDIR
L+M-BAND HR-MODE SPECTROSCOPY
Full spectral range is possible by using several blocking filter options, one blocking λ > 4 μm for 3-3.5 μm in third order
One shifting blocking filter blocking < 3.45 μm and < 3.8 μm for clear analysis of order 2 at small wavelength
OD Adjustable Parameters:
Parameters:
Detector:
Grating
ps
18 μm
#pixels
Blaze angle
2048
θ
45
#cal pixels
8
Central order
#spectral detectors
2
Central order blaze wavelength
Distance to next array
λc
?
μm
Lowest order
De
?
μm
Min. scan angle
α
Optimal wave-range
λopt,min
3,5 μm
λopt,max
5,5 μm
Scan step
5,1 μm
1
-15
o
1,45
o
90000
80000
70000
60000
50000
40000
30000
20000
10000
3
4
5
6
λ (μm)
15
3 μm
1,18
5,7 μm
9,16
λslice
R
λR
3,7 μm
50000
6 μm
Grating parameters:
a
7,21 μm
L
212,1 mm
Dcoll
150 mm
Scan range (o)
-15
Scan angle (degrees)
10
λmin
λmax
Central (slice wavelangth)
For λ
δα
Nstep
#steps
Full range
Resolution
2
Db
Performance:
o
Resolution
Pixelsize
5
0
-5
3
4
5
6
-10
-15
-20
λ (μm)
5,3
Figure 6-23: HR grating parameters for the L+M-band channel
Page 104 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
MIDIR
ELT-TRE-LEI-11200-0001
1.0
Q-BAND HR-MODE SPECTROSCOPY
Need only 4 orders to cover whole waelength range
Parameters:
OD Adjustable Parameters:
Detector:
Grating
ps
30 μm
#pixels
40000
Blaze angle
1024
θ
60
#cal pixels
8
Central order
#spectral detectors
2
Central order blaze wavelength
Distance to next array
μm
Lowest order
De
?
μm
Min. scan angle
7
α
17 μm
Scan step
λopt,max
25 μm
λmin
16 μm
13,3
λmax
27 μm
27,6
#steps
δα
Nstep
30000
20000
10000
-5,7
o
14 15 16 17 18 19 20 21 22 23 24 25 26 27
3,4
o
λ (μm)
6
20
Central (slice wavelangth)
18 μm
λslice
R
λR
25000
20 μm
Grating parameters:
a
100,8 μm
L
288,7 mm
Dcoll
144,3 mm
Scan angle (degrees)
Full range
For λ
19,4 μm
λc
?
Optimal wave-range
λopt,min
Resolution
9
Db
Performance:
o
Resolution
Pixelsize
15
10
5
0
-5
14
15
16
18
19
20
21
22
23
24
25
26
27
-10
Cross dispersor:
acrossdisp
17
λ (μm)
19,61 μm
o
Scan range ( )
-5,7
11,3
Figure 6-24: HR grating parameters for the Q-band channel
6.4.3 N-band system in detail
For implementation of the general idea, some basic design choices have been made. The
optical design must allow for the following set of drivers:
• MR and HR channels should use the same pre-optics (IFUs) and detector arrays
(significant cost reduction)
• Spectrographs should be kept compact
Preliminary tests and checks for collimation demonstrated that the required FOV for the
N-band can not be implemented with one simple optical component. For this reason, it has
been decided to move directly to TMAs for both collimator and camera. Secondly, it has
been chosen to use these TMAs in double paths, to reduce the number of optical
components and to get an easier separation of the incoming and outgoing beams in the
optical system as the collimated beams are even more expanded in size.
Figure 6-25 shows an indicative picture how the medium and high resolution
spectrometers can be configured to use the same input from the IFUs and the same FPA,
where only two selector mirrors are used to switch light from the IFUs to the different
spectrograph arms and from the spectrograph arms to the FPAs.
Page 105 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
HR-spectrograph
FPM
IFU
X-dispersion
MR-spectr.
Figure 6-25: Schematic picture of the combined optical paths for the medium and high
resolution spectroscopic modes. Figure is not on scale and indicative.
The HR spectrometer for the N-band consists of a combination of two different types of
Three Mirror Anastigmats (TMA's). The first TMA is a Cook Three-mirror objective with
intermediate focus and pupil, which uses a relatively small grating for cross-dispersion.
The second TMA is a Three-mirror Wetherall and Womble objective and has a large
echelle in its collimated beam (Figure 6-26 and Figure 6-27).
Figure 6-26: High-resolution spectrometer using TMA#1 with cross-dispersion grating
and TMA#2 with main-dispersion echelle.
Page 106 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-27: Shaded layout of N-band high-resolution spectrometer, showing 52 mm IFU
slit and 2048x2048 detector (pixels of 30 microns) in front. Cross-dispersion grating is
shown in 4 positions for 4 orders. Near intermediate focus the CdTe lens can be seen.
For pixel matching reasons the focal ratio of the system should be 2*pixel size /
wavelength. Using a pixel size of 30 microns and a wavelength λslice of 9 microns, this
results for the N-band in a working F# = 6.67. The use of two double-pass systems has the
disadvantage that the focal ratio is maintained through the whole system. The pre-optics
has to deliver the IFU slit with a F# of 6.67, with the pupil at the right position and of the
correct size. For a Cook Three-mirror objective this pupil should be located after the IFU
slit in order to be able to have a pupil at the cross-dispersion grating.
The Three-mirror Wetherall and Womble objective has its entrance pupil before its image
plane, so in principle the two systems could be coupled. However, in practice it was not
possible to match the pupils perfectly, so a CdTe field lens in the intermediate focus is
necessary to project the pupil on the echelle of the second TMA. All mirrors and gratings
are oversized by 50% to allow for the light losses caused by light diffracted by the image
slicer. The spot sizes on the 2x2 detectors (2048 pixels of 30 microns) are shown in Figure
6-28 for a central wavelength of 8.77 microns in order 12. Figure 6-29 shows the central
spot sizes for all configurations of Table 6-11.
Page 107 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-28: Spot diagrams on the detector for order 12 for three wavelengths and three
field positions. At the central wavelength of 8.77 microns the Airy disk is sampled by 5
pixels of 30 µm.
Figure 6-29: Spot diagrams on the detector for orders 12 through 9 at centre wavelength.
Page 108 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
To setup the spectrometer for a chosen wavelength, both the cross-dispersion grating and
main-dispersion echelle has to be tuned to position that wavelength in the centre of the
detector (see Table 1).
Table 6-11: Wavelengths, order and grating tilts for N-band.
config
cross-dispersion
main dispersion
140 lines/mm
16.35 lines/mm
order
tilt in
degrees
order
tilt in
degrees
λ1
λ2
λ3
1
1
37.87
12
-60.2
8.695
8.770
8.840
2
1
42.04
11
-60.2
9.486
9.567
9.644
3
1
47.45
10
-60.2
10.434
10.524
10.608
4
1
54.94
9
-60.2
11.593
11.693
11.787
A consequence of using cross-dispersion is the tilt of the spectrum direction. This tilt
reduces the usable slit length up to 25%. This can be partially corrected by using the
main-dispersion spectrometer in the quasi-Littrow "off plane" configuration, meaning that
the beam separation occurs perpendicular to the plane of dispersion. This quasi-Littrow
configuration causes an inclination of the spectral lines with respect to the spectral
direction. By choosing the correct off-plane angle this effect may nullify the tilt by the
cross-dispersion resulting in a more or less square echellogram. This only holds exactly
for one wavelength as the dispersion caused by the cross-dispersion grating changes with
wavelength. The detector has to be rotated to fit to this echellogram as good as possible.
The echellogram is shown in Figure 6-30.
Assuming a resolution of 2 pixels, the achieved spectral resolution at the central
wavelength each of order is about 54000.
Page 109 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-30: Layout of echellogram on the detector with orders 12, 11, 10 and 9 showing
different directions of spectrum for each order.
The typical component parameters of the optical design are listed in Table 6-12. In
general, the list seems very feasible apart from M4 that might exceed the size that can be
handled by traditional diamond turning (highlighted dimension).
Table 6-12: Optical elements of high-resolution spectrometer; includes 50% oversizing.
element
surface type
size (mm)
Decentre (mm)
IFU exit slit
n.a.
52
dx=0, dy=0
M0
flat
10 x 46
dx=0, dy=0
M1
even asphere
100 x 240
dx=+12, dy=0
M2
even asphere
20 x 60
dx=+28, dy=0
M3
even asphere
110 x 150
dx=+122, dy=+6
cross-dispersion grating
flat
100 x 160
dx=0, dy=0
CdTe lens
flat, spherical
10 x 80
dx=0, dy=0
M4
even asphere
260 x 590
dx=-420, dy=+20
M5
even asphere
200 x 360
dx=-155, dy=+15
M6
even asphere
260 x 400
dx=-30, dy=+24
echelle
flat
245 x 410
dx=0, dy=0
M7
flat
90 x 125
dx=0, dy=0
detector
n.a.
61.44 x 61.44
Page 110 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
6.5 OPTICAL: DESIGN MEDIUM RESOLUTION SPECTROMETER
The design of the MR spectrometer will look more or less like a hybrid of the VISIR
(RD11) and the MIRI (RD12 and RD13) design. Because of the necessity to switch
between MR and HR, a double pass collimator/camera system is the current baseline.
Table 6-13 lists the magnitude of the grating parameters required for obtaining the MR
spectrometers. The first row, labelled “Result” provides the field parameters for the
spectrometers. The dimension of the FPA is the linear scale the cameras should project the
image to. The scale, determined by the diffraction limit and spatial and spectral sampling,
is geometrically expressed by the “Camera F/#”, expressing the paraxial values of the
geometric optical system. The next rows “Collimator” are included to present some
paraxial parameters of the input of the spectroscopic camera (the grating). The F-ratio is
taken here a convenient value for the IFUs, but for a double pass system, this value has to
be changed to the required camera F-ratio. Here, an interesting row is the “Opening angle
beam fan” expressing the field angles in the collimated beam. Compressing the pupil at
the grating increases the angles on the grating. For the L+M design using 5 detectors on a
row for a FOV > 1”×1”, these angles increase to unrealistic large values. For the current
overview, a limited FOV is taken, as the IFU design needs careful checking to cope with
the large FOV. However, even for these moderate values, the angles associated with a
small collimated beam turn out to be critical in the camera design. The last rows “Grating
parameters full Littrow” show the outcome for the grating. All parameters look feasible
and are a reasonable extension of the values needed for MIRI.
Table 6-13: Paraxial grating parameters for the medium resolution spectrometer
channels. Oversizing is not taken into account.
Telescope diameter
Now implemented for the bands in MIDIR
Channel
Result
L+M
Acceptable #detectors
Obtained FOV
(arcsec)
Final number of slices/resolution elem.
Camera F/#
Collimator
Grating
Dimension FPA
Collimator F/#
(mm)
42
Q
Spec
Spatial
Spec
Spatial
Spec
2
2
2
2
2
1
0,80
1,33
1,28
1,77
1,86
0,76
Spatial
42
44
30
30
20
22
9,73
9,73
6,67
6,67
3,33
3,33
76,4
10
76,4
61,4
61,4
61,4
10
30,7
10
Total slit length @ IFU exit
(mm)
Pupil diameter
(mm)
20
20
40
40
80
80
Focal length
(mm)
200
200
400
400
800
800
Opening angle beam fan
Resolution
(degrees)
78,5
λblaze
(μm)
(B)
92,2
18,89
3000
parameters Order
full Littrow
(m)
N
92,2
11,40
3000
5,61
3000
1
1
1
3,7
9
18
Beam diameter at collimator
(mm)
Blaze angle
(degrees)
15,509
18,65
18,65
Grating constant
(μm)
6,9186
14,072
28,144
λ-range over detectors
Lgrating
(μm)
2,5259
3,072
6,144
(mm)
20,756
42,2167
84,4334
(a)
20
20
40
40
80
80
As mentioned before, finding a MR camera proved to be difficult. In principle, the HR
mode camera can be taken and scaled down a little to reduce the collimated beam
Page 111 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
diameter. This approach has been followed originally. However, to come to reasonable
collimated beam diameter including a tilt of the grating for the beam separation, the
system breaks down and the TMA can not provide sufficient optical quality any more.
Other systems were tried as well, like an upscaled MIRI design; this system becomes too
large (physical dimension of approximately 2 meters) and the Cook TMA, which does not
work because of its pupil location close to the final image plane. The main reason for
failure so far is the fact that the angles between the collimated beams in the pupil become
large and providing sufficient field separation for our FOV pushes the systems over the
limits.
As baseline, we could select a slightly smaller HR-TMA. We have to choose whether we
want to use the grating parameters of Table 6-13 on a grating with dimensions above
100mm resulting in a much higher grating resolution, or redefine a grating to give the
required spectral resolution with this collimated beam diameter. This choice is left open as
we are convinced that we can do better than found so far in this study and this action will
be taken up well before the Point Design Study will start.
6.6 OPTICAL DESIGN: CALIBRATION UNIT
In section 4.8, the constraints and components of the calibration system have been listed.
Apart from these elements, there were additional constraints place on the calibration unit
to be versatile enough to handle the atmospheric thermal background.
The calibration system consists of the following components:
•
A stable black body point source (BB-PS) with a source temperature sufficiently high
to generate enough flux and have the maximum of the Planck curve at or below 3 μm
(T > 1000 K)
•
An insertable set of polarizers (one rotatable) to be able to have full control on the
intensity from the source with a well controller output polarization
•
Tuneable monochromator (TMC)
•
An XY positioning system to be able to calibrate field positions
•
Gas Cell containing a mixture of gasses at low pressure. Temperature of this cell
should be different from source, but gasses may not condense or freeze out
•
Integrating sphere (IS) to transform the point source beam into a flat field source (high
accuracy)
•
Telescope simulator (TS), to provide the calibration signal with the proper optical
beam definition
•
Fast switch, to be able to switch (< 50 milliseconds) between science observations and
calibration checks.
Parts of the calibration system can not be placed inside the cryogenic environment.
Therefore, it is envisioned that the calibration system needs to be warm and couples its
input directly into either the pre-optics of MIDIR or the AO-system of MIDIR.
Page 112 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
For as flat field source, the current choice is to use an integrating sphere. The required
system temperature is too high to use a hot surface. This would probably lead to a too high
heat generation close to the instrument and optical paths.
Without going into detailed optical design, the functional lay-out of the calibration system
is indicated in Figure 6-31. The integrating sphere is as far down stream as possible to
prevent that filtering components distort the flat field. Details in the design of individual
components are not in this study, but can be adapted from existing instruments.
Field mask
XY-Bench
Filter wheel
BB-PS
Flux adjust
TMC
Gas Cell
TMC
Gas Cell
IS
TS
Fast
Switch
Pol filter
Figure 6-31: Block scheme of calibration unit, light from the black body source (BB-PS)
passes crossed polarizers that tune the output flux, passes a filterwheel, a tuneable
monochromator (TMC), gas cell, goes via an integrating sphere (IS) through a field mask
(on XY-table). Masking options: large field (large enough for imager) or pinhole. The
polarization filter can be used to calibrate polarization sensitivity of MIDIR. The
Telescope simulator (TS) adopts proper F-ratio, pupil location and focuses on the science
focal point via the fast switch. All units indicated with shadows can be (each individually)
switched in and out of the optical path.
The telescope simulator might be relatively big, the IS should be designed that the field is
sufficiently large for flat fielding the imager. The pinhole at this stage on an XY-table
simplifies the optical path at the cost of intensity. However, the source is planned to be
sufficiently bright to provide flat fielding levels comparable to the “brighter” astronomy
targets. The source should be actively cooled.
One issue, on sky wavelength calibration, has not been discussed here. However, with a
fast switching system a continuous operation of the calibration system is feasible, and
might even be required for the atmospheric background handling.
Page 113 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.7 DETECTORS AND FOCAL PLANE CONFIGURATIONS
Two different types of detectors will be used in the MIDIR instrument. For the L and M
band λc=2.5 μm HgCdTe Hawaii-2RG AQUARIUS 2048×2048 arrays are proposed. The
N and Q band can be covered by Si:As blocked impurity band detectors. Their
characteristics are given below. For the infrared wavefront sensor a small λc=2.5 μm
HgCdTe array may be used.
6.7.1 2K x 2K λc=5 µm Hawaii-2RG arrays
The detectors proposed for the L and M band of the MIDIR instrument are 2Kx2K
Hawaii-2RG arrays manufactured by Rockwell Scientific (see Figure 6-32). The detectors
have to be cooled to T=40K. The readout electronics will either be the Sidecar ASIC
developed by Rockwell or the NGC controller developed by ESO. In the latter case the
video signal of the detectors will be amplified with cryogenic preamplifiers located next to
the focal operating at temperatures of 70 K. The thermal gradient between detector and
preamplifier is maintained by a short flexible manganin board. The detector mount will
provide manual alignment of tip tilt and possibly a motorized focus stage. The cryostat
cabling will be made with flex boards. Each detector will be read out with 32 parallel
video channels at a maximum pixel rate of 5 MHz corresponding to a maximum frame
rate of 38 Hz. The arrays can be arranged in a close buttable mosaic. For single Hawaii2RG arrays ESO has a standard detector set-up defined in the Interface Control document
VLT-SPE-ESO-14010-3853. An overview is shown in Figure 6-33.
Figure 6-32: Mosaic of 2x2 2Kx2K λc =5 µm HgCdTe arrays.
Since ESO did not yet evaluate λc=5 μm HgCdTe Hawaii-2RG arrays but has extensive
experience with λc=2.5 μm arrays, typical performance characteristics of λc=2.5 μm
Page 114 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Hawaii-2RG arrays are summarized in Table 6-15. If the cut-off wavelength is λc=5 μm,
there is a fortuitous lattice match between the IR sensitive HgCdTe layer and the CdZnTe
substrate. Therefore, λc=5 μm arrays are expected to have equal or better performance
than λc=2.5 μm arrays.
detector mount
cold braid
connection
preamp box
flexcable part
temperature
control
Two 72 pin Micro-D
connectors (feed-through)
Vacuum
connector
128 pins
Figure 6-33: Standard ESO detector mount for single Hawaii-2r arrays
Table 6-14 Design parameters
Parameter
Units Acceptance Test
Criteria
Detector
technology
MBE HgCdTe on
CdZnTe
By design
Detector input
circuit
Source Follower per
Detector
By design
Measured
performance
Comments
Pixel pitch
µm
18
18
By design
Fill factror
%
>90
>90
By design
Spectral range
µm
0.9 to 5.0
0.9 to 5.0
By design
3 edges
3 edges
By design
Buttability
Page 115 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 6-15 Performance Parameters
Parameter
Units
Cutoff
-
Acceptance Test
Criteria
Measured
performance
5.0
5.0
Array Mean QE
e /ph
J-band QE > 0.65
K-band QE > 0.65
J -band QE = 0.69
K-band QE =
0.72
Charge storage
capacity
e-
>80000
>120000
Pixel operability
%
>95
>99.63
Array mean dark
current
e- /pix /sec <1
<0.01
Array mean read
noise (100 kHz)
e-/pix rms
< 15
21.4
Power dissipation
(100 kHz)
mW
<4
To be measured
Max bow
µm
<20
Comments
K-band,
reference
pixels
removed
CDS
Table 6-16 Cosmetics
Parameter
Units
Acceptance Test Measured
Criteria
performance
Clusters
of
100-400
contiguous bad pixels
<400
6
Clusters
of
400-4000
contiguous bad pixels
<20
4
Clusters of 4000-40000 econtiguous bad pixels
<2
0
Clusters
of
>40000 %
contiguous bad pixels
0
0
Comments
A map of the quantum efficiency in K and J band for a typical λc=2.5 µm array is shown
in Figure 6-34 and . A map of the dark current is shown in Figure 6-36 and noise map in
Figure 6-37.
Page 116 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-34: Quantum efficiency of Hawaii-2RG array #76 in K-band.
Figure 6-35: Quantum efficiency of Hawaii-2RG array #76 in J-band.
Page 117 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-36: Dark current of Hawaii-2RG array #76 in J-band.
Figure 6-37: Map of readout noise of Hawaii-2RG array #76.
A new Hawaii-4RG-15 device is now in development at Rockwell scientific. The array
has a format of 5Kx4K and a pixel pitch of 15 μm. As the Hawaii-2RG the array has
reference row and column outputs for common-mode noise rejection and a guide window
output for randomly placed guide windows, which can be read out in an interleaved way
while reading out the full science frame.
The array can be configured by software to use 1, 4, 16, 32 or 64 outputs. The multiplexer
offers the unique feature to choose between three different types of unit cell designs, the
source follower per detector design (SFD), the capacitive transimpedance amplifier
(CTIA) and the direct injection (DI). This feature allows for selecting the unit cell design
which is best optimized for the specific flux level. For best noise performance ( 9 erms)
the SFD design can be used. The storage capacity for this design is 1E5 e. The CTIA
design has a readout noise of 90 erms and a full well 9E5 and direct injection can be used
Page 118 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
for high fluxes with a readout noise of 120 erms and a full well of 3E6. The cut-off
wavelength can be tuned by changing the composition x of the Hg(1-x)CdxTe alloy between
λc=2.5 and 10 μm. With good antireflection coatings the quantum efficiency is expected to
be > 80 %.
The Hawaii-4RG array will be compatible with the SIDECAR ASIC developed for the
Hawaii-2RG array. The ASIC replaces the data acquisition system and generates all dc
voltages and clocks required to operate the array and directly digitizes the video signal on
the focal plane with 32 parallel 200 KHz ADC’s and 32 parallel 5MHz ADC’s. Only a
digital interface is required to operate the detector with the ASIC.
6.7.2 1K x1K Si:As Aquarius Arrays
For the N and Q bands the best detector choice is Si:As. At present ESO is funding the
development of a high flux 1Kx1K Si:As blocked impurity band array for ground based
applications. The Aquarius array is the long awaited replacement for the CRC-774
320x240 Si:As IBC pictured below and utilizes the 1024 x 1024 Si:As Impurity Band
Conduction (IBC) Sensor Chip Assembly (SCA ) technology developed for the JWST
MIRI instrument.
Figure 6-38: 320x240 CRC 774 Si:As array used for ground based instruments such as
Michelle, Timmi2, TRECS, VLTI-MIDI.
The basic specifications of the Aquarius array are given in Table 4. The large number of
video outputs reduces the required analog bandwidth. Each output has to read out only
2.46 Mpixels/s to achieve a frame rate of 150 Hz. The maximum possible storage capacity
is limited by the electric field of breakdown in silicon, which is 3E5 V/cm and results in a
storage capacity of 1.5E7 e- for a pixel size of 30 µm. For Nyquist sampled images the
focal ratio f# is 2d/λ with d being the pixel pitch. The flux per pixel does not depend on the
pixel size, but larger pixels allow more storage capacity, which helps to reduce readout
speed. A pixel pitch of 30 µm is the best compromise between cost of the array, storage
capacity and readout speed.
Page 119 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 6-17: basic specifications of the Aquarius 1Kx1K Si:As array.
Parameter
Units
Pixel pitch
µm
Number of video outputs
specs
30
32
Maximum frame rate
Hz
150
Storage capacity spectroscopy
e-
1E6
Storage capacity imaging
-
e
1.5E7
The readout topology of the Aquarius array is shown in Figure 6-39. The array is
organized in 2x8 stripes with the bond pads at the top and at the bottom of the array. Each
stripe has 128 columns and 512 rows. The top half of the array reads out top to bottom and
the bottom half bottom to top. The array is closely buttable in two directions and long
mosaics can be built in the direction without bond pads. Windowed readout is possible.
The frame rate scales with the number of pixels in the window.
8 or 32 outputs (selectable)
Column shift register
Row shift register
Column shift register
Figure 6-39: Readout topology of Aquarius array.
The multiplexer will be based on the VIRGO multiplexer used on the VISTA science
FPA. It will be a Source Follower per Detector design which achieves excellent noise
performance. The VIRGO multiplexer has successfully been tested at 10 K. Hence, the
risk to use this type of multiplexer for the Aquarius array is acceptable. The clocks for
Page 120 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
both the vertical and the horizontal shift registers will be accessible, which allows
implementing the read-reset-read mode. This readout mode has a duty cycle of 100% for
short integrations, whereas the reset-read-read mode, which is the only mode possible with
the VIRGO multiplexer, has only a duty cycle of 50 % for the minimum integration time.
The relative quantum efficiency and the readout noise of Si:As has been measured with
MIRI detectors by the University of Rochester as shown in Figure 6-40 and Figure 6-41.
With multiple sampling and 8 Fowler pairs the readout noise can be reduced to values as
of 10 erms, as shown in Figure 6-41.
10.0 K
MIRI Assy 7581011.1 Wafer 9601/A05
& Assy 7581009.1 Wafer 9581/A05; Diodes D28 at -1.0 V Bias
Relative Response / Photon
1.00
9/22/2004
9601 @ - 1.0 volt
hanger queen @-1.0 volt
9581 @ -1.0 volt
0.10
0.01
0.0
5.0
10.0
15.0
20.0
25.0
30.0
Wavelength (µm)
Figure 6-40: Relative quantum efficiency of MIRI detectors.
Page 121 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-41: Readout noise as function of number of Fowler sample pairs.
6.7.3 Infrared Wavefront Sensor
As a first step to develop an infrared AO sensor ESO is evaluating the CALICO detector,
which is a 128x128 pixel prototype device, which has 7 different unit cell designs. The
performance of each design can be compared and the best design will be selected for a
uniform AO sensor. At ESO first infrared images have been obtained with this device, but
the video output of the array is unstable it does not yet work as expected. By placing the
signal processing circuitry under each pixel it is possible to filter the noise prior to
multiplexing. The full exposure time is then available for limiting noise bandwidth rather
than just the pixel time. The most promising design is a two-stage capacitive
transimpedance amplifier (CTIA). The pixel pitch is 40 micron.
Rockwell Scientific is now working on the next step, the SPEEDSTER device, which is a
256x256 pixel sensor with 40 micron pixel pitch and λc=2.5 μm HgCdTe diodes. The full
frame can be read at a frame rate of 625 Hz, a 128x128 pixel window at 2.5 KHz. The
readout noise from a single read is expected to be 5 erms. The device will have 12 bit
ADC’s in the multiplexer and provide a fully digital output. With multiple sampling the
noise may be reduced to < 3erms. The storage capacity with the lowest gain will be 7E4
electrons. The multiplexer supports multiple window readout for a wide variety of
different configurations required such as Shack-Hartmann sensors, tip tilt sensors, or
fringe trackers.
Page 122 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.8 CHOPPING
6.8.1 Technical Alternatives
Observations of astronomical objects in the thermal infrared with uncompromised
sensitivity depend on signal modulation (aka chopping). The classic way used over the
years at 1-8m class telescopes, that is secondary mirror chopping, is no longer feasible at
ELTs and potentially also not really desirable. One therefore has to investigate new, or not
so new schemes. Those could be:
•
Focal Plane Chopping: The implementation scheme depicted in Figure 4-5 is not
really useful for an ELT as the stroke is extremely limited, the second beam is useless
(defocussed, blurred ...). In principle this would not matter, but it creates a big penalty,
as one has to measure the chopping offset. In the past it has also been the experience
that this scheme produces a quite high chopping offset.
Focal plane chopping, however, may be 'resurrected' if one considers moving the
detector. If that can be solved mechanically it might indeed be an extremely effective
scheme which could be used for small sources at the detection limit (small here means
an object extension less than the chopper throw).
•
Pupil Plane Chopping: This can be done by means of a relay (e.g. an Offner). In
principle this could be coupled with any adaptive optics scheme as any AO system has
a mirror somewhere to compensate for the tip-tilt error. One simply has to increase the
stroke. The associated error for the wavefront associated with the tilt of the mirror
would need to be compensated by the AO system. This means, a careful design in
order not to exceed reasonable strokes on the deformable mirror may be an excellent
way, to provide for chopping. Such a system should produce the same chopping offset
as today's M2-chopping. Because of the reasons given above, it still is not easily
possible to observe objects larger than the chopper throw.
A fundamental problem of pupil plane chopping is the interaction with active optics.
Whenever the movable mirror is ahead of the wave-front sensor this scheme needs a
“counter-chopper” which may lead to an undesirable degree of complexity.
•
Dicke Switching: This method has been used in the past in radio-astronomy. The way
to implement this in an ELT-mid/IR spectrum is given in the Figure 6-42. This method
has practically been applied very successfully for IR observations of the Sun (cf.
Deming et al. 1986 [RD15] and Glenar et al. 1988 [RD 16]). The method may not
offer the highest sensitivity, but it is most likely the only method to exploit a field of
20-30arcsec at an ELT without compromises on the spatial information resulting from
chopping (see above). This method could also be combined with a rigorous flat-field
calibration.
•
Nodding / Dithering: This method can be applied if the methods above fail. It is today
in use at telescopes, which can not provide M2-chopping or when chopping is not
possible such as in the case of Lunar occultations. The overall experience is in the US
and at ESO that the resulting frames suffer from fixed pattern noise.
Page 123 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-42: Dicke Switching, an infrared source is illuminating an integrating sphere.
The brightness can be varied smoothly and rapidly by rotating one of the two polarizers.
This source is inside of the instrument cryostat. It illuminates an integrating sphere. The
exit of the integrating sphere is re-imaged into an instrument pupil plane. By means of a
fast moderate precision kinematic mirror the instrument will observe part of the time the
sky, and part of the time the integrating sphere. Fast in this context means a transition
from one state to the other in ~50 milli-seconds. The polarizers will be aligned such, that
the resulting flux is approximately equal to the signal from the sky.
Page 124 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.8.2 Trade-Offs
The comparison of the various methods to produce the necessary image modulation for
noise filtering is tentatively summarized in Table 6-18.
Table 6-18: Summary of Signal Modulation Methods
Field
Restrictions
Focal Plane
Chopping
few arcsecs bad
0.45-0.9
technical risk
Pupil Plane
Imaging
~10 arcsec
bad
0.45-0.9
technical risk (AO
challenging,)
efficiency = 0.9 <=> AO
perfect in both beams
none
good
0.5
provides also for a very
good flat field calibration
device; should be
implemented in any case
?
good
(tbc)
Dicke Switching
Nodding/Dithering
Extended
Objects
Comments
Chopping Method
Efficiency
(exposure
time)
0.15 (– 0.9?) needs testing; will be
different for each detector
and will depend on
site/weather
needs development of
suitable fixed pattern noise
filtering
6.8.3 Recommendations and Suggestions for Prototyping
At this point one can summarize that Nodding and Dithering provide for a fall-back
solution for signal modulation and noise filtering required for a Mid-IR Background Noise
Limited Instrument. Such a scheme would in the worst case sacrifice a factor of 2.5.
However, with the next generation of detectors having less 1/f noise this might improve.
Therefore the following experiments and/or prototyping activities should be pursued with
high priority partially at the VLT UT4 once the adaptive (i.e. also non-chopping) M2 has
been installed:
•
test of nodding only performance of next generation detectors (e.g. the Raytheon
Aquarius); this will come partially from the planned upgrade of VISIR at VLT-UT3
•
in the context above, development of suitable algorithms for noise-filtering
•
test of Dicke Switching: needs an instrument prototype
Page 125 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
•
test of pupil plane chopping: needs a lab prototype and an instrument prototype
•
test of focal plane chopping by detector “wobbling”: needs laboratory prototyping
(could be done in the context of the VISIR-Aquarius upgrade) and thereafter an onthe-sky test which needs again an instrument prototype.
In conclusion, a test camera should be built; potentially this can be connected with a 2nd
generation VLT mid-IR instrument taking advantage from the active M2 retrofitting of the
VLT UT4.
6.9 CRYOSTAT CONCEPT AND TEMPERATURE REQUIREMENTS
6.9.1 Temperature Requirements
Temperature levels and heatloads:
Radiation Shield8 (120 kg):
120 K
150 W
Main Instrument Structure (1250 kg):
20 K
12 W
Detectors (16 kg):
5K
19 W
Cooldown time: < 48 hours
Cooldown of the detectors will start if the temperature of the rest of the instrument is
below 150 K in view of detector contamination.
Warm up time: < 24 hours
Temperature stability:
Main Instrument Structure: < 1K (TBD)
Detector: < 10 mK (TBC)
Vibration levels at the detector shall be limited to 3 μm.
Based on a Cassegrain location of the instrument the cooling system shall operate under
+/- 60o telescope orientation.
Cooling system shall operate at altitude > 4500 meters.
Cooling system shall have a Mean Time Between Failure better than 10,000 hrs.
6.9.2 Cooling Schemes
For an instrument like MIDIR several cooling options are feasible. In this section three in
principle rather different solutions are discussed:
8
There is no requirement on the heat shield temperature, only an optimization to minimize the overall power
requirement and benefit from the optimal efficiency of the coolers.
Page 126 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
1. Large number (about 16) of Pulse Tubes Coolers
2. Medium number (about 8) of Gifford McMahon coolers
3. Small number (1) of Helium liquefiers
Table 6-19 gives an overview of suppliers that produce coolers that deliver cooling power
in the order of Watts at 4.2 K
Table 6-19: Coolers that deliver cooling power in the order of Watts at 4.2 K.
Manufacturer
Model
Cryomech
PT410
Sumitomo
SRDK415D
Linde
L70
Helium
Liquefier
Type
Input
Specific
Power
Cooling
Power at
4.2 K (W)
(50 Hz)
Power
0.9*)
7.2
8.0
Gifford
McMahon
1.5*)
6.5
4.3
Claude
21
75
3.6
2-stage
Pulse Tube
(kW)
(kW/W)
2-stage
*)
The cooling power is based on the assumption that the cooling power
is directly at the cooler’s cold stage using standard flex lines. Actual
conditions for MIDIR i.e. strapping between cold stage and detector,
longer flex lines can give rise to performance losses.
Ad 1) Pulse Tube Coolers (PTC)
Present state of the art 2-stage PTC’s have a cooling power of about 1.5 W at 5 K. Since
the power dissipation of a detector will be about 1 W one detector will be connected to a
single PTC. Main advantage of PTC’s are no moving parts at the cold end of the cooler.
Therefore the coolers are practically maintenance free and vibration levels are about a
factor of 2 lower than for Gifford McMahon coolers. Compressor units can be connected
to the cold head by relative long flexible lines that supply an oscillating high/low pressure
Helium gas flow. Water cooling of the compressors will assure operation of the coolers at
high altitude. PTC’s are sensitive to mounting orientation. Cooling power to the
instrument will only be supplied at the location of the cold stage. Therefore thermal
strapping with associated losses will be needed to distribute cooling power to the
appropriate locations on the instrument.
Ad 2) Gifford McMahon Coolers (GM)
Present state of the art 2-stage GM-Coolers have a cooling power of about 2.5 W at 5 K.
Since the power dissipation of a detector will be about 1 W two detectors will be
connected to a single GM-cooler. GM-coolers are widely used in the semiconductor
industry and for MRI applications mainly for cryopumps. The coolers have moving
displacers at the cold end operating at about 1 Hz. Therefore vibration levels are higher
Page 127 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
and maintenance intervals are shorter than for PTC-coolers. Compressor units can be
connected to the cold head by relative long flexible lines that supply an oscillating
high/low pressure Helium gas flow. Water cooling of the compressors will assure
operation of the coolers at high altitude. GM-coolers are relative insensitive to mounting
orientation. Cooling power to the instrument will only be supplied at the location of the
cold stage. Therefore thermal strapping with associated losses will be needed to distribute
cooling power to the appropriate locations on the instrument.
Ad 3) Helium liquefier
To produce a cooling power of 19 W at 4.2 K about 28 litres/hour of liquid Helium is
needed. This volume can be produced with a relative small Helium liquefier. The liquefier
will be operated at ground level. Therefore the mass of the liquefier will not be added to
the instrument mass. The gaseous Helium that is evaporated at the detectors can be used to
cool the main instrument structure. The Helium exhaust of the instrument will be fed to
the liquefier by a closed system and will be recycled.
Cooling with liquid Helium can be implemented in two ways:
1) Liquid He-tank
Figure 6-43 gives a schematic overview of this option. A liquid He-tank will be part of the
instrument cryostat and will rotate with the instrument at the telescope. A flexible line is
connected to the instrument to lead the Helium exhaust gas to a buffer vessel that is
connected to the Helium liquefier. The liquefier will continuously produce liquid Helium.
The tank at the instrument will be refilled every 24 hours.
Figure 6-43 Liquid Helium tank as part of Instrument Cryostat
The size of the tank is roughly estimated as 1500 litres. This is based on a Helium
evaporation of about 31 litres/hour, a hold time of 24 hours and assuring that the tank can
hold 750 litres both in horizontal and vertical orientation. This volume can be packed in a
Page 128 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
cylinder with a 3 m diameter and a height of about 0.25 m. Roughly extrapolating the
(much smaller) mass of the X-Shooter NIR cryostat the mass of a 1500 litres tank will be
about 1500 kg. The telescope movements will cause sloshing of the Helium and therefore
spatial temperature variations.
For MIDIR the dissipated power at the detectors must be transported to the liquid Helium
tank over distances in the order of 1 meter. Probably conventional copper strapping can be
used since high purity copper shows very high conductivity (of about 10.000 W/m-K) at
4-5 K. An alternative solution is a cryogenic pump that will transport liquid Helium to the
detectors.
The heat capacity of the exhaust gas will give a cooling power of about 100 W at 20 K to
cool the main instrument structure. This will require a heat exchanger mounted on the
instrument structure.
2) Continuous flow system
Figure 6-44 gives a schematic overview of this option. Connected to the instrument are 2
flexible lines: a liquid Helium input line and a gaseous Helium exhaust line. Both lines are
also connected to the Helium liquefier. In this way a continuous flow system will provide
the cooling. A cryogenic pump must be used to transport the liquid Helium to the
instrument.
Figure 6-44 Continuous flow liquid Helium system
Note that liquid Helium will evaporate continuously in the liquid Helium transfer line
giving rise to significant pressure changes. The occurrence of Thermal Acoustic
Oscillations must be taken in account in the design of the transfer line. Probably the total
mass added to the instrument will be smaller than for PT –and GM-coolers.
A standard flexible liquid Helium transfer line (Cryofab) will have a heat leak of about 0.7
W/m. For a 30 m transfer line this amounts to 21 W. Therefore the total power at 4.2 K is
Page 129 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
about 60 W (including 50% margin) corresponding to a Helium consumption of 83
litres/hour. The Linde L70 Helium Liquefier will not be able to produce this amount of
Helium. However, the larger model Linde L280 Helium Liquefier is capable of producing
89 litres/hour with an input power of 160 kW.
The heat capacity of the exhaust gas will give a cooling power of about 240 W at 20 K to
cool the main instrument structure. This will require a heat exchanger mounted on the
instrument structure.
Cooling power with either liquid Helium or gaseous Helium can easily be distributed over
the instrument by piping.
6.9.2.1 Cooling Scheme Selection
The following parameters mainly determine the choice for a cooling scheme:
-
various temperature levels
cooling power range
electrical input power available
cooldown time
mass/envelope
cooler orientation
maximum operational altitude
vibration levels
temperature stability
reliability
maintenance
cost
Background information for all three cooling options concerning these parameters is given
in Section 6.9.3. Figure 6-45 shows a thermal block diagram of the instrument and gives
an overview of the expected heat flows. An overview is given in Table 6-20.
Page 130 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-45: Thermal block diagram. Expected heat loads indicated in brackets (Watts).
Table 6-20 Cooling schemes overview for detector cooling at 5 K.
Gifford-
LHe-
MacMahon
Liquefier
Cryomech
Sumitomo
Linde
PT410
SRDK-415D
L70
16
8
1
24 W @ 5 K
20 W @ 5 K
19 W @ 4.2 K
400 W @ 45 K
270 W @ 50 K
250 W @ 50 K
Electrical input power
115 kW
52 kW
75 kW
Cooldown time
Additional LN2
precooling needed
Additional LN2
precooling needed
Additional LN2
precooling needed
Mass at instrument
Coldheads: 312 kg
Coldheads: 148 kg
Mass at ground level
Compressors: 1200 kg
Compressors: 720 kg
Envelope (LxWxH) at
instrument
Per Coldhead:
Per Coldhead:
33x23x67 cm
30x14x56 cm
Pulse Tube
Type
Number of units to
provide detector
cooling
Cooling power
LHe-tank: 1500 kg
Closed sytem: - kg
3500 kg
LHe-tank: 1.5 m3
Page 131 of 204
Conceptual Design Study of MIDIR
Pulse Tube
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Gifford-
LHe-
MacMahon
Liquefier
Envelope (LxWxH) at
ground level
Per Compressor:
58x53x66 cm
Per Compressor:
50x45x69 cm
Per unit:
500x400x300 cm
Cooling water
128 LPM @ 27 oC
56 LPM @ 28 oC
72 LPM
Cooler orientation
o
< 45 -50
Max. operational
altitude
No limit
o
(max 15 % cooling
power loss)
Helium sloshing
No limit
No limit
No limit
Max 10 m/sec2
Max 0.1 m/sec2
-
Vibration levels
15 μm
26 μm
(at cold stage)
Peak to peak
Peak to peak
About +/- 200 mK
About +/- 300 mK
Helium sloshing
< 10 mK
< 10 mK
< 10 mK
Warm up of single
detector
Warm up of two
detectors
Warm up of all
detectors
Coldhead: 20.000 hrs
Coldhead: 10.000 hrs
Maintenance
Compressor: 20.000
hrs
Compressor: 20.000
hrs
8.000 hrs
Cost (kEuro)
504
308
750
Vibration levels
(at cold head)
Temp. stability
(no active control)
Temp. stability
(active control)
Failure of single unit
-
6.9.2.2 Conclusions
For an instrument mounted on a Cassegrain location at the telescope at the present
moment it will not be an option to use PTC’s because of the strong reduction of cooling
power for orientations larger than 50 degrees from vertical.
Compared to mechanical coolers cooling by liquid Helium will probably induce the lowest
vibration levels. A liquid Helium tank mounted inside the instrument cryostat will add
considerable mass (about 1500 kg) to the instrument. Therefore this option does not seem
very realistic. A closed Helium system with liquid Helium flexible transfer lines seems
feasible. However, the heat input at the transfer line will probably be larger than the
detector dissipation. The cost of a liquid Helium system is about 1.5-2 times higher than
for a cryocooler system (please note that Table 6-20 only shows the cost of the
cryocoolers for the detector cooling; for the cooling of the Optical Benches 4 additional
Page 132 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
cryocoolers are currently foreseen). In case of failure of the liquefier the complete system
will warm up.
GM coolers do not suffer from the drawbacks mentioned above. However, GM coolers are
known for introducing significant vibration levels at both the cold head and the cold stage.
Therefore these coolers can only be used in combination with either passive or active
vibration control.
In conclusion, for MIDIR mounted on a Cassegrain location at the telescope a cooling
scheme of GM-coolers with ample vibration control seems most optimal. In case of a
Nasmyth location a new trade off between the 3 cooling options must be made.
6.9.2.3 Steady State for GM-Coolers
In steady state a total of 8 GM-coolers are connected to the detectors and 4 additional GMcooler can be used to remove the heat from the main structure and the radiation shield (see
also
Table 6-21). Instead of 4 additional coolers in principle also a single cooler can be used.
However, it is preferred to have a modular instrument and for test purposes and instrument
upgrades it will be very beneficial to have cryogenic independent and self-supporting
modules that do not share the same cryogenic infrastructure. The cooling power
requirements for the 4 additional GM-coolers are roughly 5 W at 20 K and 40 W at 80 K.
Therefore the SRDK-415D coolers will not be suitable candidates.
Table 6-21 Estimated amount of GM-Coolers
# Detectors
# GM-Coolers
L+M-Spectrometer
8
4
N- Spectrometer
4
2
Q- Spectrometer
2
1
Imager
2
1
0
4
16
12
Optical Benches and
Radiation Shields
Total
A possible connection scheme for the GM-Coolers, i.e. option 2, is indicated in Figure
6-46. The actual temperature levels are based on a balance of the heatloads of the various
levels and the available temperature dependent cooling power of the coolers.
Page 133 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-46 Connection scheme for option 2 (GM coolers)
6.9.2.4 Cooldown for GM-Coolers
For MIDIR the distribution of the masses connected to the first and second stages will be
as follows: the relative low mass of the detectors will be connected to the second stages
and the large mass of the main structure will be connected to the first stages. However,
above 77 K the 2nd stage of the coolers will be connected to the Optical Bench for two
reasons:
1) The temperature of the detectors will be kept high during cooldown of the
Optical Bench in view of contamination.
2) During cooldown most of the cooling power is needed at the large mass of the
Optical Bench.
Below 77 K the 2nd stages of the GM-coolers will be disconnected from the Optical Bench
and connected to the detectors. At this point the cooldown of the detectors will start.
Practically, this can be implemented with two types of heat switches 1) Heat switches that
are ON above about 77 K, and 2) Heat switches that are ON below about 77K.
A sketch of the temperature profile in time is given in Figure 6-47:. In the rough
calculations elements are only implemented as heat capacity. Additional heat resistance, as
for instance the cooldown of large optical elements, will result in larger cooldown times.
Page 134 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-47: Cooldown profile for option 2 (GM coolers)
Page 135 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.9.3 Background Information
6.9.3.1 Various Temperature Levels
The detector temperature is determined by the temperature dependent detector
performance. The instrument temperature is determined by the requirement of background
limited observations, the cold shield temperature can be chosen to optimize the power
consumption taking into account the radiative losses and the efficiencies of the coolers.
6.9.3.2 Cooling Power Range
Dissipation
The number of detectors and the expected power dissipation at 5 K are indicated in Table
6-22. A dissipation of 1 W is assumed for a single detector.
Table 6-22: Overview of detector power dissipation.
# Detectors Power dissipation
(W)
L+M-Spectrometer
8
8
N- Spectrometer
4
4
Q- Spectrometer
2
2
Imager
2
2
Total
16
16
Conduction Supports
The heat-load for the supports of the OB is calculated by an empirical scaling equation
[RD 10]. The material for the supports is assumed to be Titanium:
C = αKm 0.66 ΔT = 0.022 × 5e − 2 × 1250 0.66 × (300 − 20) = 34 W
where
C Heatload (W ), α = 0.022 (high stress),
K (heat conduction W / cm − K ), m (mass kg ), ΔT
(temp.difference K )
The length of the supports will be about 0.3 m. Combined with the heatload of 34 W this
gives a total cross-section of 7200 mm2.
To minimize the heatload on the 20 K structure a heatsink of the supports on the Radiation
Shield is foreseen.
Page 136 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Conduction Harness
The heatload for the instrument harness is mainly determined by the copper cross-section
that is needed to warm up the instrument within the specified warm up time. For the
harness a length of 1 m is assumed.
To warm up 1250 kg of Aluminium from 20 to 300 K within 24 hours requires a power
input of about 2500 W. If the electrical power is supplied by a relative low (and safe)
voltage of 50 V this yields a current of 50 A.
The main concern in current carrying wires is the occurrence of hotspots in the middle of
the wires. The most critical situation arises at warm up if the hot and cold sides of the wire
are at room temperature. To limit the maximum temperature in the middle of the wire to
about 50 oC and taking into account supply and return wiring the total cross-section
amounts to 40 mm2.
The cross-section of the harness for a single detector is assumed to be about 2 mm2. For a
total of 16 detectors this amounts to 32 mm2.
To minimize the heatload on the 20 K structure a heatsink of the harness on the Radiation
Shield is foreseen.
Radiation Shield
To minimize the radiation heatload from ambient on the cryostat the use of Multi Layer
Insulation (MLI) is assumed. The following equation developed by Lockheed Martin (C.
Keller et al, Final Report: Thermal performance of Multilayer Insulations, NASA
Contractor Report Number CR-134477, 1974) is used
C N 3.56 2
Cε
q= 1 s
T − Tc2 + r tr Th4.67 − Tc4.67
2 N +1 h
Ns
s
(
)
(
)
where q is the heatload in W/m2, Th is the temperature of the hot side, Tc the temperature
of the cold side, N is 30 is the layer density in layers/cm, Ns is 20 is the total numbers of
layers, εtr is 0.031 the room temperature surface emissivity, Cs Emperical contstant with
numerical value 2.11×10-9, Cr Emperical constant with numerical value 5.39×10-10.
This equation predicts a heatload of about 1 W/m2 for a temperature gradient from 300 K
to 77 K. However, the equation is valid for relatively ideal blankets with no seams or
penetrations. Therefore the equation is multiplied by a degradation factor of 5. The total
outside surface area of the cryostat will be about 19 m2. Therefore the expected heatload at
the radiation shield is about 95 W.
Page 137 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Radiation window
The diameter of the entrance window is assumed to be about 100 mm. The radiation
heatload can be estimated from
R = εσA(Th4 − TL4 ) = 1 × σ × (7.85e − 3) × (300 4 − 20 4 ) = 4 W
where
R
Heatload
(W ), ε
Emissivity , σ
A Surface area window (m 2 ), TH
TL
Temperature cold
Boltzmann cons tan t ,
Temperature hot
side ( K )
side ( K )
Table 6-23 gives an overview of materials and dimensions of the thermal conductors.
Table 6-23: Thermal conductors.
C1
C2a
C2b
C3
H1
H2
H3
Description
Material
X-section (mm2)
Length (m)
Radiation Shield
Support
G10
9250
0.25
Titanium
7200
0.2
Titanium
7200
0.1
Epoxy
600
0.03
Copper (RRR=100)
40
1
Copper (RRR=100)
40
1
Copper (RRR=100)
32
0.25
OB Support
Ambient-Heatsink
OB Support
Heatsink-OB
Detector support
Harness
Ambient-Heatsink
Harness
Heatsink-OB
Detector Harness
6.9.3.3 Electrical Input Power
The estimated electrical input power for the 3 options is indicated in Table 6-20.
6.9.3.4 Cooldown Time
A large VLT-instrument like CRIRES with a mass of 550 kg will be cooled down to a
temperature of 65 K in about 30 hours (ESO Messenger, No. 114, December 2003).
Page 138 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Therefore a cooldown time specification for MIDIR with a mass of 1250 kg to 20 K in
about 48 hours seems reasonable.
Figure 6-48 shows a typical cooldown for a crycooler (in this case a Sumitomo RDK408D). Masses of 18 kg and 90 kg Copper are connected to the first and second stage
respectively. The mass of 90 kg is cooled down to 20 K in about 55 hours.
Figure 6-48 Cooldown behaviour of Sumitomo RDK-408 GM-Cooler (data provided by
Sumitomo)
From Figure 6-48 the mass of Aluminium that can be cooled down to 20 K in 48 hours is
estimated as 2 × 12 × 90 x (48/55) × (79/170) = 880 kg where the factor of 2 assumes that
the cooling power of the first and second stages are identical during cooldown, the factor
of 12 is the number of coolers and the factor (79/170) takes into account the difference in
heat capacity for copper and aluminium.
Since the mass of the Cold Bench is estimated as 1250 kg it can be concluded that precooling is needed. In practice it is relatively cheap and easy to use liquid Nitrogen for the
pre-cooling. For the cooldown of 1 kg of Aluminum roughly 1 litre of liquid Nitrogen is
needed. Therefore it takes roughly 500 litres of liquid Nitrogen to speed up the cool down
of the Optical Bench.
6.9.3.5 Mass/Envelope
Mass and envelope estimates of the cooling systems at the instrument and at ground level
are indicated in Table 6-20. The mass estimate for the PT- and GM-coolers at ground level
Page 139 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
is based on the assumption of one compressor for each cooler. In practice a single
compressor for multiple coldheads will be more efficient.
6.9.3.6 Cooler Orientation
The cooling power of mechanical coolers tends to have a relation with the mounting
orientation.
35
30
nd
0.8 W on the 2 stage
25
st
1 stage keeps at 45 K
20
5.0
15
4.5
4.0
10
nd
2 stage
temperature
nd
2 stage temperature (K)
5.5
0
10
20
30
5
40
50
60
st
st
1 stage capacity
1 stage capacity (W@45K)
40
6.0
0
0
Offset from vertical position ( )
Figure 6-49 Cooling behaviour of PT410 PT-Cooler as function of orientation (data
provided by Cryomech)
Figure 6-49 shows the cooling power for the PT410 Pulse Tube Cooler as function of
orientation. It can be concluded that this cooler can not be used in an orientation larger
than 50o from vertical. Therefore Pulse Tube Coolers are no option for the cooling of
MIDIR at a Cassegrain mounting on the ELT.
In principle the Sumitomo GM coolers can be mounted in any orientation. The maximum
reduction in cooling power is 15 %.
In the case of a liquid Helium tank mounted to an instrument at a Cassegrain location the
telescope movements will cause Helium sloshing and therefore temperature variations.
Page 140 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.9.3.7 Maximum Operational Altitude
For operation at an altitude over 4500 meters water cooling instead of air cooling for the
compressors will be needed. In general the cooling water will be used in a closed circuit
where the water is cooled with the ambient air. Therefore the decrease of the air density
with altitude must be taken into account.
Possibly special measures for the heat removal of motors that are used for Helium
liquefication are needed.
6.9.3.8 Vibration Levels
In general mechanical coolers will induce vibrations both at the cold head and at the cold
stage (see Figure 6-50).
Figure 6-50 Vertical acceleration and displacements for the cold head and cold stage
(from T. Tomaru et al, Cryogenics 44 (2004) 309-317) for a GM and a PT Cooler. Note
that this data is for a Sumitomo PTC instead of the Cryomech PTC that is considered in
this section.
Page 141 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Measured accelerations for the coldhead are about two orders of magnitude larger for GMcoolers (about 10 m/sec2) than for PT-coolers (about 0.1 m/sec2). On the other hand the
peak to peak displacement is only about a factor of 2 larger for GM-coolers (26 μm) than
for PT-coolers (15 μm).
It is evident that mechanical coolers can only be used in combination with vibration
control, either passive or active.
Boiling of liquid 4-Helium at 4.2 K can be quite violent and therefore induce vibrations.
At the present stage it will not be possible to make quantitative statements about the
induced vibration levels at the instrument. It can be an option to pump away the Helium
vapour and reduce the pressure above the liquid Helium to about 50 mbar. This will lower
the boiling point to 2.2 K. At this temperature the Helium will go through a change of
state from He-I to He-II and the violent boiling behaviour will stop due to the high thermal
conductivity of He-II.
6.9.3.9 Temperature Stability
The required temperature stability for the detectors is 10 mK (TBC). Mechanical coolers
in general will show periodic temperature variations corresponding to the frequency of the
moving parts. The temperature stability for Pulse Tube Coolers is about +/- 200 mK and
for GM Coolers about +/- 300 mK.
It is common practice to reduce the temperature variations to better than +/- 10 mK by
active temperature control consisting of a temperature sensor, a heater and a temperature
controller.
In the case of a liquid Helium tank mounted to an instrument at a Cassegrain location the
telescope movements will cause Helium sloshing and therefore temperature variations.
Probably also active temperature stabilization will be needed.
6.9.3.10 Reliability and Failure
Cooler or liquefier reliability can be described by Mean Time Between Failure (MTBF).
For the coolers under study here no values for the MTBF have been found.
Each detector will be coupled to a singe PTC. Failure of a cooler will lead to the warm up
of only one detector.
Two detectors will be coupled to a single GM-cooler. Failure of a cooler will lead to the
warm up of two detectors.
All detectors and the main instrument structure will be cooled with Helium. Therefore
failure of the liquefier will lead to the warm up of the complete instrument.
6.9.3.11 Maintenance
Included in Table 6-20 is an overview of the maintenance intervals for the coolers as
specified by the suppliers.
Page 142 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.9.3.12 Cost
Table 6-24 gives an overview of the purchase cost for the various coolers. Also indicated
in the table is the input electrical power for a single unit. The input power will probably
dominate the operational costs.
Table 6-24 Cost of various coolers as indicated by suppliers.
Manufacturer
Model
Cooling
Power at
4.2 K
(W) (50
Hz)
Input
Price
Cost
Power
(kEuro)
(kEuro/
W)
(kW)
Cryomech
PT410
0.9*)
7.2
31.5
35.0
Sumitomo
SRDK415D
1.5*)
6.5
38.5
25.7
Linde
L70
Helium
Liquefier
21
75
750
35.7
*)
The cooling power is based on the assumption that the cooling power is
directly at the cooler’s cold stage using standard flex lines. Actual
conditions for MIDIR i.e. strapping between cold stage and detector,
longer flex lines can give rise to performance losses.
**)
Liquefier Coldbox, Compressor, Oil Removal and Gas Management and
Control System
6.10 MECHANICAL SETUP AND METROLOGY SYSTEM
The size and the multiple unit instruments will be defined by special design, material and
thermal control.
6.10.1 General Considerations
The mechanical design for MIDIR is driven by a number of general considerations and
choices. The most important ones are:
•
General layout of the sub-units and operational orientations
•
Consequences of cryo-vacuum instrument environment
•
Material choices
•
Mechanisms, active control
•
Modularity
Page 143 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Before presenting a baseline proposal for the packaging and mechanical design of MIDIR
we discuss these topics briefly in sections 6.10.1.1to 6.10.1.5.
6.10.1.1 General Layout of the Sub-units and Operational Orientations.
The MIDIR instrument consists of the following seven optical subsystems:
•
Common pre-optics
•
Adaptive optics
•
Calibration unit
•
Imager and low-resolution spectrometer
•
3 medium/high resolution (MHR) spectrometers for the wavelength channels L+M,
N and Q.
The general layout of these subsystems is sketched in Figure 6-51. The logical sequence of
the light paths between these systems naturally puts important constraints on the layout
and packaging of the mechanical design.
Cal.
unit
cryostat
Imager
LR
spectro
Fore
optics
AO
Spectro
1
Spectro
2
Spectro
3
Figure 6-51: Layout of the subsystems
The range of operational orientations is an important additional input requirement for the
mechanical design, in particular with respect to the tolerances on flexure and stability. In
this study we assume that MIDIR will be mounted in a (quasi)-Cassegrain focus on an altazimuth telescope.
Page 144 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.10.1.2 Consequences of Cryo-Vacuum Instrument Environment
In order to achieve sky background limited sensitivity (‘BLIP’) performance, the detectors
and most of the instrument optics have to be cooled to a very low temperature. At spectral
resolutions of R~50000 the system temperature should be below ~20 K; the detectors
require operational temperatures of about 5-7 K. Most of the MIDIR subsystems should
therefore be mounted inside one or more cryostats under high vacuum conditions. This
requirement has many practical consequences; in particular it asks for:
a) a design that is as much as possible temperature-invariant
b) cryogenic mechanisms
c) a compact configuration in order to reduce the size and mass of the cryostat(s)
d) light-weighting of mechanical and optical components in order to reduce thermal
timescales.
The first of these design goals automatically leads to a strong preference for all-reflective
optics and for a homogeneous design, with optical elements and structures made out of the
same material. An all-reflective optical design is clearly desirable for a mid-infrared
instrument anyway.
6.10.1.3 Material Choices
In addition to the usual material requirements (strength, elasticity, specific weight,
manufacturability, cost), the cryogenic nature of thermal infrared instruments puts extra
requirements on the thermal material properties (CTE, conductivity). The goal of a
homogeneous design implies that the chosen material should be suitable for the production
of accurate optical surfaces. The following table gives an overview of the most critical
properties for a number of possible material types.
Table 6-25: Some global material properties
Material
stiffness
stiffness/
mass
Strength
strength/
mass
CTE
Thermal
Conductivity
Optics
possible
cost
Steel
high
moderate
moderate
/ high
moderate
moderate
moderate
not likely
low
Invar types
moderate
moderate
moderate
moderate
low
low
not likely
moderate/
high
Aluminium
moderate
High
moderate
high
high
moderate
yes, proven
low
Special
aluminium
types
(RSP,
low CTE ..)
moderate
High
high
(very) high
moderate
/ high
high
yes,
in
developm.
moderate/
high
SiC types
very
high
(very) high
high
high
very low
low
yes,
in
developm.
high
Epoxy Carbon
composites
very
high
(very) high
high
high
very low
low
yes,
in
developm.
high
hard alloy
Page 145 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Although various alternative options deserve further investigation, our present baseline
material choice for the cooled optics and mechanics is classical hard aluminium alloy (e.g.
6061 or 5083). For the cryostat(s) and warm main support frame we assume stainless steel.
This choice is based on the existing experience with other mid-IR instruments
(MICHELLE, VISIR, MIDI, MIRI).
6.10.1.4 Mechanisms, Active Control
MIDIR will require cryogenic mechanisms for three types of functions:
a) Selection of optical components or switching between optical paths
b) Adjustment of grating angles
c) Active alignment adjustment (flexure compensation).
Although the goal is to minimize the number of cryo-mechanisms, a number of
mechanisms for beam switching and optics selection are unavoidable. In principle all of
these movements can be done by rotations around single axes, but suitable cryogenic
linear actuators for alternative solutions exist already. Tilt adjustments of the gratings
require high angular resolution and stability (at the level of 0.1-1 arcsec) but the rotation
ranges are small in this case.
The need for mechanisms in category c) is not yet fully clear, but in view of the typical
dimensions and weight foreseen for MIDIR (see below) we expect that some form of
active flexure compensation will be needed. Such control systems will probably involve
small adjustments in more than one coordinate, i.e. combinations of rotations and
translations. This could require new types of cryo-actuators.
Cryogenic mechanisms have the reputation of being difficult and expensive, but this
technology is developing rapidly. Next to the traditional DC or stepping motor types,
promising new cryo-mechanisms based on piezo actuators are appearing. The latter are
very interesting for cryogenic applications due to their small size, simplicity and low cost.
New stepper-like piezo developments combine linear strokes of >100 mm and high
resolution (10-100 nm) with dissipation-free automatic locking. Simple stack piezo’s
provide small strokes (10-100 micron), large forces with high resolution (nanometers). At
the same time various new cryogenic encoders are being developed. Cryogenic rotation
encoders with resolution down to ~1 arcsec are already common. Linear encoders are still
more difficult and costly, but also here there is rapid development. Simple inexpensive
linear encoders with ~5 micron resolution are available already; higher resolutions should
be feasible in the near future.
Various kinds of new cryogenic motors and encoders will be mature within the timeframe
of the MIDIR development. They will make it possible to consider a wider application of
mechanisms for control of active or even adaptive optical elements in cryogenic
conditions. We expect that this will not only broaden the possibilities to move/adjust
optical components and detectors, but that it will improve the calibration capabilities of
the next generation of infrared instruments with respect to the traditional ‘static stability’
design philosophy. Decisions about the optimum balance between passive stability and
active control can only be made after more detailed analysis in the next phase of this
study, but the design of the instrument metrology, both hardware and software, could
become an important part of the instrument development.
Page 146 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.10.1.5 Modularity
After investigating several opto-mechanical layout options we have chosen a modular
approach, with multiple small cryostats for the main subsystems. The primary reasons for
this choice are:
a) The modularity of a multiple cryostat design gives more freedom for instrument
development in phases, in line with the likely step-wise development of the
telescope.
b) Modular packaging can result in a more compact configuration with short optical
and thermal paths.
c) A single cryostat around the whole instrument would become very large, heavy
and difficult to handle.
d) A single cryostat also requires a large and heavy internal support structure for
stable mounting of the combined cold optics. The thermal response time would
become very long. A design with multiple smaller cryostats can make use of a stiff
central support structure that is part of the vacuum enclosure but not part of the
cold mass.
Page 147 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.10.2 The Baseline Mechanical Design
The mechanical design that we adopted as baseline for MIDIR is illustrated in the
following figures.
Central
structure
Optics
Imager
Optical bench
Vacuum
vessel
Figure 6-52: Baseline mechanical design for MIDIR. Top panel: exploded view of the
central support structure and two of the three subsystem cryostats. Below: layout of the
subsystem optical paths.
The optical designs of the imager, the spectrometer pre-optics, and the N-band HR
spectrometer were used to dimension the system and work out the packaging. It was
assumed that the optical designs of the Q and LM-band spectrometer are of similar size as
the N-band system. This is not an unrealistic assumption as the relative FOV of the
spectrometer increases with decreasing wavelengths, counteracting the expected size
Page 148 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
reduction for short wavelength instruments. The imager and the pre-optics for the
spectrometer are packed very closely together directly at the output of the AO module,
conform figure 6-8. Light going to the spectrometer channels are dichroically separated
and directed to the inputs of the corresponding IFU optics of each spectrometer arm. The
IFUs are small boxes (not drawn) directly at the input focus of each spectrometer.
The mechanical structure consists of four main components:
1) A central stiff triangular cryostat which contains the pre-optics, the AO system
and the imager. This cryostat acts as the central support structure for the entire
instrument; its top flange is the mounting interface to the telescope and holds the
cryostat entrance window. The calibration unit is mounted on the warm side of this
top flange. Since high stiffness and high eigen-frequencies are important design
drivers for this central structure, stainless steel is a logical material choice. The
resulting CTE differences with respect to other (aluminium) instrument
components may require some active compensation.
2) Three smaller cryostats for the three MHR spectrometer arms. These three
spectrometer modules are mounted onto the three side faces of the central
structure. The four cryostats - the three outer modules and the central one - have
their own closed cycle coolers, but they are coupled to each other via the flanges of
the central structure. They thus share a common vacuum.
The modularity of the MHR spectrometers allows the possibility to operate incomplete
MIDIR configurations with one or more MHR units removed. Naturally an unused ‘open’
side of the central cryostat needs to be closed by a vacuum flange in that case, but it is
possible to make an interface that allows (dis)mounting of a sub-system while maintaining
the cryogenic condition in the rest of the system. Although the four cryostats have a
common vacuum, they are thermally rather well decoupled. In principle it is therefore
possible to apply different operational temperatures for the four compartments, but the
possible (dis)advantages of individual temperature regimes need to be investigated.
One mechanical advantage of this modular design is the fact the individual cold optics
units can be attached to the stiff central frame via relatively short mounting rods. This
makes it easier to achieve stable isostatic mountings with low weight.
Additional baseline design choices:
a) Cold reflective optics: all-aluminium, highly light-weighted, with optical surfaces
(gold coated) and mounting structures integrated as much as possible into single
components.
b) In general the optics mountings should be non-adjustable (i.e. alignment by
design+manufacturing precision) but the need for compensation of instrument
flexure by active control of specific components should be investigated.
c) Instrument structure: all-aluminium, highly light-weighted.
Page 149 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 6-53 More detailed views of the instrument packaging in different projections.
Page 150 of 204
Conceptual Design Study of MIDIR
All modules attached
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
One MHR cryostat removed
One MHR instrument removed, flange Top view, without central structure
closed
Figure 6-54 Modularity of the MHR spectrometer units.
Page 151 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
6.10.3 Size and Mass Estimates
For the baseline design outlined in section 6.10.2 we have made the size and mass
estimates summarized below in Figure 6-55 and Table 6-26.
Figure 6-55 Size estimates for the baseline mechanical design.
Table 6-26: Mass estimates for the baseline mechanical design.
weight
Volume Weight instrument
only [kg] Comments
Unit
[m3]
[kg]
Central cryost.
1,4
664
106 relative highly packed
spectro cryost.1
2,3
784
126
spectro cryost.2
2,3
784
126
spectro cryost.3
2,3
784
126
Backbone
692
111 steel construction
Electronic racks total
1000
Total estimated weight
4700
4 separate units
Page 152 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
7 Instrument Sensitivities and Comparisons
In this chapter we estimate the sensitivities of imager and spectrographs and compare
them to other current and future facilities. All estimates assume point sources (see section
7.4 for extended sources). The assumption of a diffraction limited point source is clearly a
“best case” scenario, and often not achieved from the ground. However, it is a well
defined situation and the most interesting science cases for MIDIR have structures close to
the diffraction limit of the telescope.
7.1 ASSUMPTIONS AND CALCULATIONS
The sensitivity of MIDIR has been estimated for all modes and wavelengths. This section
describes the assumptions that went into the calculations and presents the results. In the
following discussion we will mainly concentrate on spectroscopy.
The performance of MIDIR has been estimated for two sites:
•
Paranal Observatory at 2600m altitude, and
•
Chajnantor plateau at 5100m altitude,
and assuming in both cases mid latitude, winter time, and observations near zenith. For
these conditions we calculated atmospheric transmission and emission9 with HITRAN-PC
for two atmospheric resolutions:
• R=3000
• R=50000.
The sensitivities were then calculated at the maximum spectral resolution using a rather
complex EXCEL spreadsheet10. The fixed, wavelength-independent input parameters are
listed in Table 7-1.
Table 7-1 Fixed parameters used for the (spectrograph) sensitivity calculations.
Parameter
Value
Comment
Telescope primary mirror diameter
42 m 30 m – 60 m are considered
Secondary mirror obscuration
6.5 m
Teff of the atmosphere on Chajnantor
235 (251)K winter (summer)
Teff of the atmosphere on Paranal
245 (255)K winter (summer)
Teff of the telescope on Chajnantor
250 (267)K winter (summer)
Teff of the telescope on Paranal
262 (279)K winter (summer)
9
HITRAN-PC computes the effective atmospheric emission in units of [W cm-2 wave#1 sr-1]. To convert
this unit to [W cm-2 μm-1 sr-1] the emission has been multiplied by 104 λ-2[μm].
10
A copy of the EXCEL file can be obtained directly from Bernhard Brandl (brandl@strw.leidenuniv.nl).
Page 153 of 204
Conceptual Design Study of MIDIR
Parameter
Doc. No
Issue
Value
ELT-TRE-LEI-11200-0001
1.0
Comment
Emissivity of mirrors M1 & M2
0.04
Emissivity of gaps in M1
0.05
Emissivity of central obscuration
0.00 requires effective blocking
Emissivity of telescope spiders
0.01
Emissivity from dust contamination
0.005
Reflectivity per telescope mirror
0.98
Total number of warm telescope mirrors
2
Number of internal cold reflective surfaces
Reflectivity of individual cold surface
16
0.99
Grating efficiency
0.7 per grating (2 in high res.)
3 – 20% varies with λ
IFU slicing losses
Window & dichroic transmission
0.6 two dichroics
Spatial pixels per resolution element
2
Spectral pixels per resolution element
2
0.5 – 0.8 increases from 3.5 – 7.5 μm
Strehl ratio (AO correction)
f/# at the detector
9.73,6.67,3.33 LM, N, Q band, respectively
Physical detector pixel size
30μm
6e- for 10 non-destructive reads
Read noise per frame
10e-/s
Dark current
Detector quantum efficiency (average)
0.7 maximum at 15μm, λ dependent
Time lost due to overheads
0.8
The detector integration times (DITs) have been chosen to minimize the number of reads
while staying below the full well capacity of 1×106e– for the AQUARIUS chip. From our
calculations we derive the maximum exposure times in Table 7-2.
Table 7-2: Maximum exposure times per waveband and spectrograph mode. The
integration times at L band could be even longer (≥60s) but have been limited to 10s to
allow for off-line de-rotation in software.
Module DIT at R=3000 DIT at R=50000
LM
10s
10s
N
1s
10s
Q
0.1s
1s
Page 154 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
The following discussion illustrates how the sensitivities were calculated:
The minimum detectable flux density from a source, Ssrc [photons s-1 cm-2 μm-1], is given
by:
S src
N tot
11
2. The total noise , which depends on the background noise per pixel and the number of
pixels npix to be combined from each resolution element in both the spectral and the
1. The signal-to-noise ratio σ S / N ≡
spatial direction: N tot = N back n pix .
3. The detected signal Sel in [e-], which depends on the source flux density Ssrc, the
integration time tint, the effective collecting area of the telescope Atel, the throughput of
the atmosphere ηatm, the total throughput of telescope and instrument ηtot, the Strehl
ratio SR, the detector responsivity ηDG, and the width of the resolution element Δλ :
S el = S src SRΔλAtelη D Gη atmη tot t int
Combining (1.) – (3.) yields:
S src =
σ S / N N back n pix
S el
=
SRΔλAtelη D Gη atmη tot t int SRΔλAtelη D Gη atmη tot t int
Before we calculate the resulting detection threshold for a given S/N we need to discuss
the other quantities that enter the above equation:
The total system throughput ηtot without the atmosphere, and (currently) constant with
wavelength is the product of:
•
•
•
•
•
the total reflectivity of all telescope mirrors ηT
the total reflectivity of all instrument mirrors ηI
the fractional slice transmission ηfst. As the PSF grows with wavelength the
relative slice width gets narrower and diffraction broadening of the beam leads to
light losses. Therefore the medium resolution spectrograph will consist of several
modules. We assume a slice width of λmin/D with a fractional slice transmission of
80% at the nominal design wavelengths of λLM = 3.7μm, λN = 9.0μm, λQ = 18μm
decreasing with wavelength across the band.
the transmission of filters and dichroics ηdic, and
the grating efficiency ηg.
Hence:
ηtot = ηTη Iη fstη dicη g
11
We only consider noise from the background here; shot noise from the source signal is neglected.
Page 155 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
The total background intensity [W cm-2 sr-1 μm-1] at the focal plane is the sum of the
contributions from the warm telescope BT and the atmosphere BA times (approximately)
the total system throughput ηtot :
Btot = (BT + B A )η tot
where the background signal from the warm telescope assumes black-body emission:
⎡
⎤
2hc 2 ⎢ ε
⎥
BT = 5
λ ⎢ k hcTλ ⎥
⎢⎣ e B − 1⎥⎦
and BA as provided by the HITRAN calculations.
The total signal Sback per pixel [e– /s] from the background is the number of electrons that
are being generated solely by the background flux every second in a detector pixel. It is
the product of total background intensity Btot , the pixel “field of view” A×Ω, the width Δλ
of a resolution element, and a conversion factor, which relates the photons of energy hc/λ
to a given “light power” [W] for a given wavelength and detector responsivity ηDG:
S back = Btot ⋅ AΩ ⋅
η D Gλ
hc
⋅ Δλ
The A×Ω product at the detector plane in [cm2 steradians] is the field-of-view over which
each detector pixel sees the background. The f-number at the detector used for the
computation is determined by design considerations (to sample the slice width with two
pixels), namely 9.7, 6.7, and 3.3, for the LM, N, and Q band, respectively.
⎛
⎛
⎛ 1
AΩ = 2π ⎜1 − cos⎜⎜ arctan⎜⎜
⎜
⎝ 2# D
⎝
⎝
⎞ ⎞ ⎞⎟ 2
⎟⎟ ⎟ D pix
⎟
⎠ ⎠ ⎟⎠
The total noise per pixel for a given integration time tint is a combination of three
components (assuming Poissonian error distributions for three statistically independent
components):
1. the noise associated with the background signal: √(Sback × tint),
2. the noise associated with the detector dark current: √(Id × tint), and
3. the detector read noise and the number of frames: Nread √n
Hence, the total background noise [e– /pixel/tint] is: N back =
2
S back t int + I d t int + N read
n
With the above equations we can now compute the two most important quantities: the
flux an unresolved line Sline and the continuum sensitivity Scont.
The line flux Sline in [W m-2] can be derived from the minimum detectable signal from a
source Ssrc in [photons s-1 cm-2 μm-1] via: S line =
hc
λ
S src Δλ ⋅10 4
Page 156 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
⎡ W ⎤
− 26 c
, the continuum sensitivity Scont [Jy] can then be
With S λ ⎢ 2
⎥ = Sν [Jy ] ⋅ 10
λ2
⎣ m μm ⎦
calculated to: S cont =
hc
λ
S src ⋅ 10 4 ⋅
λ2
c
⋅ 10 26 = hλS src ⋅ 10 30
Continuum and line sensitivity are the two quantities that are plotted in section 7.3.
7.2 IMAGER SENSITIVITY
Table 7-3 lists the background fluxes expected for broadband imaging. Given the high
background fluxes, it is clear that the most gain from an ELT can be achieved for point
sources, unless one accepts significant pixel resampling (see section 7.4). For these
numbers, Figure 7-1 shows the imager point source sensitivities as a function of
wavelength for three telescope apertures.
Table 7-3: Background flux for broad band applications. ‘N-pixel’ means Nyquistsampling at diffraction limited resolution..
Atmospheric BG
BG
band
[mag/arcsec2] [JY/arcsec2]
BG[γ/s/arcsec2] BG
For a 42m tel.
[γ/s/N-pixel]
J
16.5
0.39 10-4
2.06 10+06
7.8 10+1
H
14.4
1.74 10-3
1.20 10+07
7.9 10+2
Ks
13.0
4.15 10-3
3.83 10+07
4.5 10+3
L
3.9
7.96 10+0
1.19 10+11
3.6 10+7
M
1.2
5.40 10+1
1.15 10+12
7.2 10+8
N
-2
2.51 10+2
5.03 10+12
2.5 10+10
Q
-6
2.61 10+3
3.80 10+13
2.1 10+12
7.3 SPECTROGRAPH SENSITIVITY
7.3.1 Performance of the R=3000 medium resolution spectrograph
Figure 7-2 to Figure 7-7 show the 10-σ LM, N, and Q band point-source continuum and
line sensitivities for a 42m telescope on Chajnantor in winter time.
Page 157 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
Point source sensitivity (10s , 1h, R=5)
1,00E+00
Limiting flux [mJy].
1,00E-01
1,00E-02
1,00E-03
1,00E-04
1,00E-05
0
5
10
15
20
25
30
Wavelength [µm]
continuum sensitivity [mJy]
Figure 7-1: 10-σ point source sensitivity for a 30m, 42m and 60m ELT in one hour
integration time.
1.000
0.100
0.010
0.001
3.0
3.5
4.0
4.5
5.0
5.5
wavelength [um]
Figure 7-2: 10-σ, 1hr continuum sensitivity to a point source at R = 3000 in LM-band for
DIT=10s.
Page 158 of 204
line sensitivity [W m-2]
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
1.0E-19
1.0E-20
1.0E-21
3.0
3.5
4.0
4.5
5.0
5.5
wavelength [um]
continuum sensitivity [mJy]
Figure 7-3 10-σ, 1hr line sensitivity to a point source at R = 3000 in LM-band for
DIT=10s.
10.0
1.0
0.1
7
8
9
10
11
12
13
14
wavelength [um]
Figure 7-4: 10-σ, 1hr continuum sensitivity to a point source at R = 3000 in N-band for
DIT=1s.
Page 159 of 204
Doc. No
Issue
line sensitivity [W m-2]
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
1.0E-18
1.0E-19
1.0E-20
7
8
9
10
11
12
13
14
wavelength [um]
continuum sensitivity [mJy]
Figure 7-5 10-σ, 1hr line sensitivity to a point source at R = 3000 in N-band for DIT=1s.
1000.00
100.00
10.00
1.00
0.10
17
19
21
23
25
27
29
wavelength [um]
Figure 7-6: 10-σ, 1hr continuum sensitivity to a point source at R = 3000 in Q-band for
DIT=0.1s.
Page 160 of 204
line sensitivity [W m-2]
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
1.0E-18
1.0E-19
1.0E-20
17
19
21
23
25
27
29
wavelength [um]
Figure 7-7 10-σ, 1hr line sensitivity to a point source at R = 3000 in Q-band for
DIT=0.1s.
7.3.2 Performance of the R=50,000 (25,000) High Resolution Spectrograph
At very high resolution the main interest is usually not in the continuum but in the
sensitivity to narrow spectral features. Since R=50,000 is too high to be plotted for the
entire band we show here the sensitivities for three exemplary regions in the LM, N and Q
bands, which contain important diagnostic lines, namely:
•
•
•
CO (ν=1-0) band at 4.7μm (M-band)
H2 S(3) at 9.6649μm (N-band)
[S III] at 18.7130μm (Q-band)
Figure 7-8 to Figure 7-10 show the 10-σ point-source line sensitivities at a resolution of
R=50,000 (25,000 for Q band) in units of [Wm-2].
The Q-band transmission suffers from several narrow opaque regions and the actual line
sensitivity depends can only be estimated accurately on a case-by-case basis. However,
numerous important spectral features, such as the H2 S(1) line or the [S III] fine structure
line, fall into windows of good atmospheric transmission and can be observed at
unsurpassed sensitivity.
Page 161 of 204
line sensitivity [W m-2]
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
5.E-21
4.E-21
3.E-21
2.E-21
1.E-21
0.E+00
4.60
4.65
4.70
4.75
4.80
wavelength [um]
line sensitivity [W m-2]
Figure 7-8: 10-σ, 1hr point-source sensitivity to the unresolved CO (ν=1-0) 4.7 vibrationrotation band transition at R = 50000 and DIT=10s.
5.E-20
4.E-20
3.E-20
2.E-20
1.E-20
0.E+00
9.55
9.60
9.65
9.70
9.75
wavelength [um]
Figure 7-9: 10-σ, 1hr point-source sensitivity to the unresolved H2 (0,0) S(3) 9.6649μm
line at R = 50000 and DIT=10s. Obviously, the exact sensitivity depends here largely on
the exact central wavelength, including Doppler shifts.
Page 162 of 204
line sensitivity [W m-2]
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
7.E-20
6.E-20
5.E-20
4.E-20
3.E-20
2.E-20
18.60
18.65
18.70
18.75
18.80
wavelength [um]
Figure 7-10: 10-σ, 1hr point-source sensitivity to an unresolved [S III] 18.7130μm line at
R = 25000 and DIT=1s.
7.4 EXTENDED SOURCE SENSITIVITY
Most sensitivity estimates here and in other documents are based on point sources,
observed at the diffraction-limit of the telescope. It is important to keep in mind that the
angular diameter of a diffraction limited source shrinks linearly with increasing telescope
aperture D – relative to the extended background – and thus the sensitivity increases
approximately with D2. To achieve a certain signal-to-noise (S/N), the required observing
time scales with the telescope diameter as D-4.
However, if the pixel scale is always matched to the diffraction limit of the telescope,
bigger telescopes will only provide the same sensitivity per pixel to extended emission
than smaller telescopes do. If the instrument provides a sufficiently large field of view
and only the larger scales of extended features are of interest, pixel resampling will gain
sensitivity. Resampling the pixel scale to the diffraction limit of a smaller telescope will
improve the sensitivity linearly with D, and the observing time needed to achieve a certain
S/N drops with D-2 in this case. This is another reason for using an IFU in the
spectrograph design.
7.5 OTHER MID-IR FACILITIES (CURRENT AND FUTURE)
Spitzer has been successfully launched in August 2003. With its new, large format MIR
arrays Spitzer is orders of magnitudes more sensitive than its predecessors and opened up
a new observing space, discovering hundred thousands of new infrared sources in our
Galaxy as well as at higher redshifts. During its five year cryogenic lifetime Spitzer will
deliver many exciting results, but also an extremely rich dataset for high resolution followup observations with MIDIR and the facilities listed below. The catalogues that are being
compiled from the Galactic GLIMPSE survey, the extragalactic, deep GOODS, the wide-
Page 163 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
field SWIRE survey and other large/Legacy projects will provide an extremely rich
database for many years to come, just as the IRAS catalogues have been serving infrared
astronomers for more than 20 years. The Japanese satellite ASTRO-F and other survey
missions like WISE (see below) will provide further catalogues of infrared sources across
the entire sky.
7.5.1 Mid Infrared Instrumentation on 8m-class Telescopes
Table 7-4 compares the current generation of mid-IR instruments on 8m-class telescopes.
Basically all large telescopes in operation are offering a mid-IR instrument that combines
direct imaging and spectroscopic capabilities. It is interesting to see that VISIR at the VLT
is the only MIR that offers imaging pixel scales well beyond the nominal diffraction limit
sampling.
Table 7-4: Observational capabilities of MIR-instrumentations at existing 8m-class
telescopes.
Telescope
Instrument
Waveleng
th
coverage
[µm]
Gemini N
Michelle
7 - 26
Pixel scale Detector
[arcsec/pixel size
]
0.10
320x240
Spec. Res.
Window
200@7-14
KBr
110@16-26
1000@7-26
3000@7-26
30000@7-26
Gemini S
T-ReCS
8-26
0.10
320x240
100@10
80@20
1000@10
GTC
CanariCam
8-26
0.08
320x240
KBr, ZnSe,
KRS-5
(real time)
150@10,20
ZnSe,
1300@10
KBr,
900@20
KRS-5
(real time)
HobbyEberly
Keck
-
-
-
-
-
-
LWS
3.5-25
0.085
128x128
270@10
KBr+ZnSe
540@20
4000@10
4000@20
LBT
-
-
-
-
-
-
Page 164 of 204
Conceptual Design Study of MIDIR
Subaru
Comics
8-26
0.13
Doc. No
Issue
320-240
ELT-TRE-LEI-11200-0001
1.0
250,2500,10000
@10
ZnSe, KBr
2500 @ 20
VLT
Visir
8-24
0.075
2x256x25
6
0.127
0.2
350@10
3200@10
25000@10
7.5.2 Expected Contemporaries of MIDIR
7.5.2.1 The James Webb Space Telescope/MIRI
JWST will be launched in 2013 (status July 2006) from an expandable launch vehicle into
an orbit at the L2 Lagrange point. An operational life time of at least 5 years is planned,
possibly elongated to a maximum of 10years (cooling exclusively provided by closed
cycle cooler). Thus, even for the longest assumable operational lifetime JWST will
probably have ended its operation when an ELT becomes fully operational. Nevertheless,
JWST with MIRI will be a main competitor for MIDIR. Therefore, its observational
capabilities as well as some mission information are presented here in some more detail:
James Webb Space Telescope
Spectrometer Optics
JWST
• D, ~6,5m (>25m2)
• NIRSPEC
• L2, T<50K
• MIRI (50%)
FPM
• Launch Aug 2011
• 5...10 years
• NIRCAM, NIRSPEC, MIRI
• NASA, ESA, CSA
Input-Optics and
Calibration
Imager
FPM
Image Slicers
FPM
Deck
CFRP Hexapod
Figure 7-11: Left: JWST (Artist’s impression); Right: JWST-MIRI Optical Module.
JWST – Scientific Objectives:
1. The End of the Dark Age: First Light and Re-ionization
2. The Assembly of Galaxies
3. The Birth of Stars and Protoplanetary Systems
4. Planetary Systems and Origins of Life
Page 165 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Level 1 Baseline Science Requirements (Mission Success Criteria):
1) Measure the space density of galaxies to a 2 µm flux density limit of 1.0 x 10-34 Wm2
Hz-1 via imagery within the 0.6 to 27 µm spectral band to enable the determination of
how density varies as a function of their age and evolutionary state.
2) Measure the spectra of at least 2500 galaxies with spectral resolutions of
approximately 100 (over 0.6 to 5 µm) and 1000 (over 1 to 5 µm) and to a 2 µm
emission line flux limit of 5.2 x 10-22Wm-2 to enable determination of their redshift,
metallicity, star formation rate, and ionization state of the intergalactic medium.
3) Measure the physical and chemical properties of young stellar objects, circumstellar
debris disks, extra-solar giant planets, and Solar System objects via spectroscopy, and
imagery within the 0.6 to 27 µm spectral band to enable the determination of how
planetary systems form and evolve.
JWST Science Requirements:
JWST should be capable of/provide:
•
Wavelength range 0.6 to 27 µm
•
Imaging (3<R<200) with ≥16 discrete filters over 0.6 < λ < 27 µm
•
Coronographic imaging capabilities over 2 < λ < 27 µm
•
Spectroscopy with 50 < R < 5000 over 0.6 < λ < 27 µm
•
Primary mirror, unobscured ≥ 25 m2
•
Diffraction limited imaging at λ = 2 µm
•
Sensitivity:
1) 1.2 x 10-34 Wm-2,Hz-1, SN=10, 104s, R=4 (NIRCAM)
2) 1.2 x 10-33 Wm-2,Hz-1, SN=10, 104s, R~100 (NIRSPEC)
3) 7 x 10-33 Wm-2,Hz-1, SN=10, 104s, R=5 (MIRI)
•
ZL background limited imaging over 0.6 < λ < 10 µm
•
Calibration accuracy: imaging 5%, coronographic imaging 15%, spectroscopy 15%
•
FOV [arc min2] > 3.5 (MIRI)
•
Observing anywhere within celestial sphere, over 1 year
•
> 35% of sphere accessible anytime
•
Mission lifetime ≥ 5 year (propellant for 10 yr)
Page 166 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 7-5 Sensitivity estimates for the MIRI Imager and Spectrometer
Imager
Band (µm)
Bandpass
(µm)
Estimated
Background (e/s)
Detection Limit (µJy)
(10-σ in 104 sec)
EOL
BOL
5.6
1.2
6
0.19
0.15
7.7
2.2
45
0.28
0.23
10
2
94
0.7
0.5
11.3
0.8
52
1.7
1.15
12.8
2.5
222
1.4
0.9
15
4
526
1.8
1.1
18
3
672
4.3
3.1
21
5
2354
7.3
5.7
25.5
3.9
7677
29
25
Spectrometer
Wavelength
(µm)
λ/Δλ
Estimated
Background (e/s)
Detection Limit
(10-20 Wm-2)
(10-σ in 104 sec)
6.4
2400
0.04
1.2
0.8
9.2
2400
0.08
1
0.75
14.5
1600
0.5
1.2
0.8
22.5
1200
3.5
5.6
5
7.5.2.2 SAFIR
Safir (the Single Aperture Far InfraRed Observatory) is a large (10m-class) cold (4-10K)
space telescope for wavelengths between 20 micron and 1mm. This project has been
selected for a Vision Mission study currently being performed by the NASA centres
GSFC, JPL, MSFC and JSC in collaboration with Ball Aerospace, Lockheed-Martin and
Northup-Grumman. With a wavelength region of 20-800µm, SAFIR is poised to bridge
the spectrum between JWST and ALMA, improving the point source sensitivity compared
to Herschel and/or Spitzer by up to three decades (see Fig. 65).
Safir is planned to be launched on a Delta IV-H rocket at middle JWST lifetime in 20152020. Several concepts are currently under discussion to optimize the deployable telescope
assembly and the corresponding cooling concept. Main telescope requirements are:
•
Aperture diameter >8m
•
Temperature ~ 4K
Page 167 of 204
Conceptual Design Study of MIDIR
•
Wavelength range 20 – 800µm
•
Diffraction limited for λλ > 40µm (1 arcsec)
•
Pointing accuracy 0.5 – 1 arcsec
•
Pointing stability ~ 0.1 arcsec
•
Lifetime > 5 years
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
SAFIR will be injected into a quasi-stable L2 halo orbit. Currently planned instruments:
•
Camera for the 20-600µm wavelength region
•
Low resolution spectrometer for the wavelength region 20-800µm (R~100)
•
Mid resolution spectrometer for the entire wavelength region (20-800µm) R~2000
•
High resolution spectrometer for the 25 to 520µm region (R~106)
Following these specifications, SAFIR could be a very powerful complementary facility to
future MIR instrumentations like MIDIR and JWST/MIRI. Within the small overlapping
wavelength range between 20 and 27µm, compared to ALMA, SAFIR provides by a
factor of 10 higher sensitivity for point sources but a factor of at least 10 lower spatial
resolution.
7.5.2.3 ALMA
Following the current time-scale ALMA will detect first light using the full array in 2010.
The baseline frequency bands available are: 86 – 116GHz, 211 – 275 Ghz, 275 – 370 Ghz,
and 602 – 720GHz. Spatial resolution can be changed between 350 arcsec/Freq[GHz] and
4.2arcsec/ freq[GHz] depending on the chosen configuration. Thus, the maximum spatial
resolution is achieved at 86GHz (417µm) with 6 mas.
7.5.2.4 Darwin
The European Space Agency has selected the "InfraRed Space Interferometer - Darwin" as
a mission for its Horizons 2000 programme. Selection of a launch date, probably at or
after 2015, will be made on cost, science and technology grounds sometime before then.
Darwin will use a flotilla of three space telescopes, each at least 3 metres in diameter, and
a fourth spacecraft to server as communications hub. The telescopes will operate together
to scan the nearby Universe, looking for signs of life on Earth-like planets. This is a
daunting challenge and will require a number of technological innovations before the
mission launches in the middle of the next decade.
7.5.2.5 VLTI
At MIR wavelength, VLTI is in operation already since few years (MIDI). As
interferometric device, it can not be competitive to a 30m to 60m telescope in sensitivity,
however, the spatial resolution is of the same order as MIDIR or even better.
Page 168 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
7.5.2.6 WISE
WISE will provide an all-sky survey from 3.5 to 23 microns up to 1000 times more
sensitive than the IRAS survey. WISE has been selected by NASA as its next MediumClass Explorer. With this decision the WISE mission will proceed into Phase B
(Definition Phase).
A 40cm-telescope is collecting the light into four channels to produce a complete four
colour survey of the sky at 3.3, 4.7, 12 and 24 µm, which will provide an excellent data
base for JWST and other MIR pointing facilities like MIDIR. Thus, WISE is not
competitive to JWST and/or MIDIR, neither in resolution nor in sensitivity, but will
provide a detailed study of selected astrophysical objects.
7.6 PERFORMANCE COMPARISONS
In the previous section we have seen what
However, as we will show in this section, MIDIR does not only fill very important niches
in the parameter space, but is also very competitive even with future space facilities,
where the wavelength ranges overlap. Table 7-6 gives a summary of the most relevant
instrument parameters.
Table 7-6: Comparison of the main mid-IR “competitors” of MIDIR.
Project
Wavelength
range [µm]
Telescope
diameter
[m]
Telescope
temperature
Diffraction
FOV
limit @5µm
[arcmin]
[mas]
Launch
JWST (MIRI)
5 – 28
6.5
50 K
159
2.3 x 2.3
2013
MIDIR
1 – 27
30-60
290 K
34-17
1x1
2015
SAFIR
30 – 500
10
5K
100
Spitzer IRAC
3.6 – 8
0.85
70 K
1213
5x5
2003
WISE
3.5 – 24
0.40
15K
2580
45 x 45
2008
2020
Page 169 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 7-12 shows a comparison of the point source continuum sensitivity between
MIDIR and several other infrared and sub-millimeter missions. Although MIDIR cannot
surpass JWST-MIRI in imaging sensitivity it will provide much higher angular resolution,
depending on the E-ELT aperture (Figure 7-13).
1,00E-02
Spitzer
Herschel
JWST
1,00E-03
SAFIR
ALMA
30m ELT
1,00E-04
42m ELT
60m ELT
Limiting Flux [Jy]
1,00E-05
1,00E-06
1,00E-07
1,00E-08
1,00E-09
1,00E-10
1,00E-11
1,00
10,00
100,00
1000,00
Wavelength [µm]
Figure 7-12: Comparison of point source sensitivity of contemporary IR and submillimeter instruments to MIDIR on a 30/42/60 m ELT.
Resolution [arcsec]
100
Spitzer
Herschel
JWST
SAFIR
30m ELT
42m ELT
60m ELT
ALMA high
ALMA low
Angular resolution [arcsec]
10
1
0,1
0,01
0,001
1,00
10,00
100,00
1000,00
Wavelength [µm]
Figure 7-13: Comparison of spatial resolution of contemporary IR to sub-millimeter
projects to MIR-instrumentation at an ELT.
Page 170 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 7-14 compares the sensitivities between Spitzer, JWST and MIDIR for medium
resolution spectroscopy. Due to the relatively lower background noise MIDIR is
performing extremely well in this comparison, approaching the JWST-MIRI sensitivity
but at higher angular resolution. At even higher spectral resolution (R=50,000) MIDIR
will be unsurpassed.
Spitzer (R=600)
JWST-MIRI (R=2400)
MICHELLE (R=3600)
MIDIR (R=3000)
line sensitivity 10sigma, 1hr
[1E-19 W/m2]
1000.00
100.00
10.00
1.00
0.10
0.01
5
7
9
11
13
15
17
19
21
23
wavelength [um]
Figure 7-14 Point source line sensitivity comparison between MIDIR on a 42m E-ELT,
JWST-MIRI, Gemini-Michelle, and Spitzer-IRS for an unresolved line detected at 10-σ in
one hour.
Figure 7-15 illustrates the huge gain in the parameter space of spatial and spectral
resolution that MIDIR will provide.
We conclude that most of the currently planned projects will be complementary to MIDIR,
either in wavelength or in spatial resolution: SAFIR and ALMA will work at longer
wavelengths, JWST and WISE will not have the resolution of MIDIR, and VLTI will not
provide sufficient sensitivity for most of the science cases for MIDIR. In particular the
comparison with JWST-MIRI reveals several areas (highest angular resolution, medium-,
and high resolution spectroscopy) where MIDIR will comparable or even superior to
MIRI (while MIRI will be unsurpassed for studies of extended or outside the atmospheric
bands).
In summary, MIDIR could be expected to fulfil the need for a highly sensitive and flexible
mid-infrared instrument providing highest spatial resolution over a long period.
Page 171 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Spectral resolution with spatial resolution @3-30µm
100000
E-ELT
JWST
1000
Spitzer
100
Spectral resolution power
10000
10
10
1
0,1
0,01
1
0,001
Spatial Resolution [arcsec]
Figure 7-15: Comparison of the areas in the parameter space of spectral versus spatial
resolution covered by Spitzer, JWST-MIRI and MIDIR.
Page 172 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
8 Specific MIDIR Requirements on the Telescope
8.1 REQUIREMENTS ON THE TELESCOPE SITE
The feasibility of most science case and the competitiveness of MIDIR with JWST/MIRI
depends to a large extent on the properties of the atmosphere at the telescope site. At M
and Q-band the performance is mainly limited by the atmospheric transmission, while at
N-band the performance is mainly given by the temperature of atmosphere and telescope,
although the effective width of the N band depends on the transmission. The transmission
properties of the atmosphere will determine if unique, important diagnostics (such as CO
at 4.7µm or the H2 line at 17.03µm) are accessible. At the long wavelength part of the Mband the sensitivity from a site like Chajnantor is about one order of magnitude better than
from Paranal (Figure 8-1). A very significant factor is the amount of precipitable water
vapour. The magnitude and timescales of its fluctuations require more study since
fluctuations may become a strong component of the image degradation – despite AO
correction – and may require wavefront/tip-tilt sensing at N-band. In any case, the
amplitude of such an effect is expected to be much reduced at high altitudes.
Figure 8-1 Comparison between a 42m telescope/MIDIR spectrograph on Chajnantor
(blue) and Paranal (red). The better atmospheric transmission at the higher site will yield
a gain in sensitivity of about one order of magnitude longward of 5μm.
8.2 REQUIREMENTS ON THE TELESCOPE FOCUS
In this section we will discuss the advantages and disadvantages of several telescope focus
positions for the performance and operation of MIDIR. We will base our discussion on the
currently two leading (out of five) telescope designs discussed by the ELT Science &
Engineering Working Group (ESE-WG). Figure 8-2 shows those two telescope concepts
(both based on an aspherical primary mirror):
Page 173 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
•
•
ELT-TRE-LEI-11200-0001
1.0
a Richey-Chrétien design (RC)
a five mirror design (5M)
M2
M1 –
36000
M2
=
M4
M5
M1
M3
892.86
Scale:
0.0028
CM
24-Apr-06
Figure 8-2: The two possible telescope designs. Shown on the left is the Richey-Chrétien
design and on the right the five mirror design.
The RC design offers two possible foci: a classical Cassegrain focus (removing the fold
mirror M3 for MIDIR operation), and one of the “standard” Nasmyth platform foci. The
5M design could be seen as a quasi-conventional Ritchey-Chrétien solution with an
intermediate pseudo-Cassegrain focus (above the primary) followed by a 3 mirrors
Nasmyth AO module. Hence this telescope design also offers two possible foci for
MIDIR. The ESE-WG notes that, “should an adaptive secondary mirror be considered
realistically feasible while still in the design phase, it would be possible to remove the 3
mirrors module and transform the 5 mirror solution in a conventional 2/3 mirrors
Cassegrain/Nasmyth telescope”. However, an adaptive secondary mirror for high order
wavefront control is currently not part of the telescope baseline. Table 8-1 compares the
parameters of the two 42m telescope designs, both using an F/1 primary mirror.
Page 174 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 8-1: Comparison of the main technical differences between the two leading
telescope concept designs.
Ritchey-Chretien
Concept
Secondary mirror
Other mirrors
Nasmyth
F/Number
• 2 mirrors Ritchey
Chretien
• Flat folding for Nasmyth
4.5m Convex
•
M3 : Flat/4m
16
5 mirror Nasmyth
• 2 mirrors Ritchey Chretien
• relay optics to Nasmyth
6m/convex
•
•
•
M3:Cv/4.2m
M4:Flat/2.6m
M5:Flat/2.8m
15.9
4.5
(intermediate)
angular
FOV
/Linear
Obstruction (area)
Baffling
Field stabilization
10 arcmin/2m
10 arcmin/1.944m
1%
10%
No baffling
Baffling in relay optics
M2
M5
Although providing a fully AO corrected beam on a gravity stable platform, the main
disadvantage of the 5M Nasmyth focus is the high thermal emissivity expected from the
additional three warm mirrors. In order to maintain excellent performance over a long
time, an optimized coating is not sufficient and frequent cleaning and/or recoating are
likely to be required. It is a major concern that the regular mirror optimization, which is a
delicate operation with the active AO elements, may not happen frequently enough for
best MIDIR performance. In addition, the baffling foreseen for the 3-mirror AO element
may add to the thermal background.
The 5M Cassegrain focus offers the smallest thermal background level of all options – a
tremendous advantage that overcompensates its disadvantages: the limited accessibility –
in particular since MIDIR will need an additional AO system to be tested and
commissioned at that location –, a rather fast beam, and a non-gravity stable focus. An
open issue to be addressed by the telescope design working group is the change between
these two foci and the attached instruments.
Both foci in the RC design offer a reduced thermal background from the telescope based
on only three mirrors. Although the RC Cassegrain design is listed with only two mirrors
in the ESE-WG document, the large focal length of M2 will provide a focus far behind the
primary mirror, and may require an additional beam folding mirror.
In Table 8-2 we compare the various advantages and disadvantages of the four potential
focus positions for MIDIR.
Page 175 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 8-2: Comparison of the potential focus positions regarding optimal MIDIR
performance.
RC Cassegrain RC Nasmyth 5M Cassegrain 5M Nasmyth
F/# number
+
+
o
+
Gravity stability
-
+
-
+
Pupil rotation
+
-
+
-
Emissivity
o
o
+++
---
AO required
-
-
-
+
Accessibility
+
+
o
+
Total
+
+
++
o
Generally, the large size and mass of the instrument, its high opto-mechanical stability
requirements over long times, and the source tracking requirements at centi-arcsecond
level, recommend an active image control by a telemetry system. However, such a system
will be able to compensate for flexure due to a changing gravity vector, which is therefore
not considered a problem for the two Cassegrain foci.
To some extend, the MIDIR imager and spectrograph optics are independent of the
telescope f/#number because of the additional MIDIR AO system (except for the 5M
Nasmyth focus). The input beam to both MIDIR components – and hence the output of the
MIDIR AO system – is f/10. The AO system will be designed to work with either an input
beam of f/4.5 (5M Cass) or f/16 (all others).
In summary, the favoured focus position for MIDIR is provided by the 5M Cassegrain
focus. The least attractive option is the 5M Nasmyth focus. The two foci provided by the
RC telescope design are acceptable although not optimal.
8.3 REQUIREMENTS ON THE TELESCOPE PERFORMANCE
The performance of any Mid-IR Imager and Spectrograph will critically depend on the
thermal background emission from telescope+AO system. Hence, a telescope with the
minimum number of warm surfaces is clearly preferred. The IR-optimized reflectivity of
each surface – which includes both the initial surface coating and the dust-free
preservation of the surface – is of crucial importance. Figure 8-3 shows that a 30 meter
telescope with only two optimized mirrors (2% emissivity per surface) will yield the same
sensitivity at N-band as a twice as large, 42m telescope with five “normal” mirrors (5%
emissivity, each).
Page 176 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Figure 8-3 Sensitivity comparison between an N-band spectrograph on an IR-optimized,
two mirror, 30m telescope, and a non-IR-optimized, 42m telescope with five mirrors. The
performance in terms of sensitivity is essentially the same.
8.4 FIELD- AND PUPIL ROTATION
8.4.1 Instrumental De-rotation
Mechanical de-rotation is best suitable for instrumentations of large FOVs with compact
optical design. The whole instrument is rotated around the optical axis. As MIDIR will be
moving with the telescope, flexure problems due to changing direction of gravity should
be compensated by some (slow) TT-mechanism. The same TT-loop could compensate for
flexure effects due to the mechanical rotation. As mentioned above, this de-rotation mode
introduces the need of counter-rotating the pupil stop.
8.4.2 Detector De-rotation
De-rotation by rotating the detector in general is a solution, too, but should be avoided for
IR-detector arrays due to their high sensitivity to EMC-effects, to changing wiring and
thermal coupling problems.
In addition, detector de-rotation can be applied in imaging mode only. Thus, this is not a
solution for MIDIR.
8.4.3 Optical De-rotation
Optical de-rotation is best suitable for small FOV diameters. The largest advantage is the
small amount of weight that has to be rotated, flexure effects can be kept small and no
cable twister is required. These advantages are paid by additional optical components that
Page 177 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
might be much larger than the FOV. For MIR-instrumentations these components
contribute additional thermal background if they are not included into the cryostat.
If high accuracy polarimetry is required, de-rotation should be done by non-optical
methods: The large tip-angle required for compact de-rotators are producing serious
instrumental polarization effects.
8.4.4 De-rotation by Post-processing
De-rotation by post-processing is only possible if single integration time, the FOV and the
zenith distance are small enough such that the rotation near the corner or the FOV is small
compared to the pixel pitch. Single integration time at TIR and MIR wavelengths in
general are short enough in this sense (see below). However, derotation by post-processing
increases the required data flow: In general, for MIR ground based observations it is not
necessary to store the individual frames, single DITs can be co-added on-line, only the
mean value and standard deviation are stored. If de-rotation by post-processing is applied
– if not provided on-line –, the individual frames have to be stored, a drastically increase
of the data flow is the consequence. Nevertheless, we favourite here de-rotation by postprocessing.
The maximum (meridian) velocity of field rotation is given by :
w = dp/dt =w0 cos(F)cos (A) /cos(a)
where p is the paralactic angle, F is the observatories Latitude, A the azimuth (measured
westwards from the south-point), a is the altitude (measured zenith-wards from the
horizon) and w0 is the sidereal rotation rate.
w0 = 15 arcsec/s
The maximum (meridian) velocity of field rotation is given by:
wmax =w0 cos(F)/sin(δ-F)
Page 178 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
10
maximum image rotation (h=0) [ w 0]
8
6
4
2
0
-90
-80
-70
-60
-50
-40
-30
-20
-10
0
10
20
30
40
50
60
70
-2
-4
-6
-8
-10
Declination (deg)
Figure 8-4: Maximum (at meridian) field rotation versus declination for Paranal. ω0 = 15
deg/hour
Maximum single integration time (DIT) will be 60ms (see VISIR ITC). Assuming that for
a 2048x2048 array at the edges the rotation during DIT should be smaller than 0.1pixel at
any time, the maximum allowed field rotation without application of an image de-rotator
is then 14.24 arcsec/60ms, that is 15.8ω0. For Paranal this means that Zenith distance
should not be smaller than 3.3 deg.
In case of the IFU, integration times may be significant longer, the FOV, however is much
smaller. Assuming a FOV of 128x128 pixels, the single integration times can be larger by
a factor of 16, thus, DITs up to 1s are acceptable down to a zenith distance of 3.5deg
without de-rotation.
As long as the instrument is fixed to the E-ELT focal station, there is no pupil rotation.
Image de-rotation should not be provided by rotating the whole instrument or by optical
parts in front of the Lyot-stop, as in this case counter-rotation of the pupil-mask becomes
necessary.
8.5 SUITABILITY OF MIDIR AS A “FIRST LIGHT” INSTRUMENT
The conceptual design presented in this document is reflecting the considerations of
MIDIR being a first-light E-ELT instrument. MIDIR would provide diffraction-limited
images at 10μm over a large field of view at about the same angular resolution as JWST in
the near-IR and HST at optical wavelengths. Hence, the data from MIDIR by themselves
Page 179 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
or in combination with diffraction limited data at shorter wavelengths from HST, JWST,
VLT would provide stunning images of significant PR value.
MIDIR may also provide important technical information during the late commissioning
phase of the telescope structure, alignment and emission from the thermal-IR perspective.
Last but not least, the requirements on the wavefront quality are much relaxed at mid-IR
wavelengths. Co-phasing of the mirror segments, windshake, AO complexity, and
alignment errors may all contribute to an overall wavefront error too large to reach the
diffraction limit at optical/near-IR wavelengths during the first period of E-ELT operation.
The option for an early commissioning of MIDIR has been taken into account by its
modular design. The early optical separation of the instrument modes and the mechanical
setup of several cryogenic modules around a warm support structure allow for a gradual
increase in complexity if necessary. The modularity allows for assembly and extended
testing already at subcomponent level, which makes the parallel development of the
instrument modules very efficient. Possible problems in one channel will not affect other
channels/modes or prohibit the use of the instrument at the telescope. The high degree of
automation allows to a large extend self-optimization and continuous testing &
verification at subcomponent level. Altogether, we conclude that MIDIR is very well
suited to be a first-light instrument at the E-ELT.
Page 180 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
9 Management
The main drivers in projects like MIDIR are the top-level requirements, quality, schedule
and money. On top of this is the risks assessment. Complex system developments like
ELTs with a variety of instruments, broad technologies and large consortia have to rely on
a low risk approach in combination with maximum scientific return. This also applies to
MIDIR. It does not mean that the design has to be conservative, but that it is designed in
such way that the main goals can be reached without unpredictable risks in budget,
schedule and performance.
This approach affects the conceptual design of the instrument. For example: a phased
approach can provide low risk instrument modules early on, while higher risk modules can
be added at a later stage. A modular approach also clearly defines the interfaces and
permits regular upgrades to instrument modules with relatively low impact on the science
operations. Hence, this is the approach we followed with MIDIR. Other early design
principles are material choices, an efficient calibration scheme, basic design principles on
production, active or passive stiffness, and active or passive thermal expansion
compensation. Some of these trade-offs require further study.
In this chapter we provide a budget estimate, a predicted schedule, and a discussion of risk
items associated with the baseline concept. We emphasize the importance of first class
project management to efficiently coordinate the work on such a complex instrument with
its many interfaces and international partners. The site of the E-ELT and its operational
constraints may also affect the requirements on instrument reliability and the quality
assurance procedures applied during its construction. Hence, it is clear that our estimates
have to be rather uncertain at the present time and should only be seen as a “best guess”
rather than an accurate cost breakdown. In particular it is important to note the following:
The budget and schedule estimates are based on the instrument as defined by the science
requirements. No budget or schedule limitations have been taken into account for the
design of MIDIR as described in this report. Our main emphasis has been put on scientific
needs and technical feasibility. Trade-offs between complexity and scientific performance
as well as possible savings due to more standardized or innovative approaches, which
might lead to a cheaper instrument, are subject to a detailed follow-up study.
Page 181 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
9.1 BUDGET
9.1.1 Introduction
Table 9-1 shows an overall cost estimate for the MIDIR project, including both hardware
and manpower but listed separately. We have also split the cost estimates for the MIDIR
instrument from its related AO system to allow for a better comparison with other E-ELT
instruments.
We have considered two independent approaches: One was to compose the list of
hardware components (as listed in section 9.3) and associated man power. The other one
was to look at several existing or currently being developed instruments for 10m class
telescopes and to scale them to MIDIR based on its complexity in relation to other
instruments and the estimated project duration and the size of the team. The estimates
derived in these two ways were in relatively good agreement with each other. However, it
should be noted that:
•
The budget estimate does not include a correction for inflation.
•
The budget estimate does not include contingencies. Generally, contingency is mostly
required in manpower estimates. Short delays can easily become more expensive than
hardware.
•
The cost items are not limited to the interior of the instrument but include also the
interfaces to the telescope, and extended software control and auto-calibration.
•
The budget estimate, for completeness, includes all instrument aspects, even the ones
which are usually covered by ESO and do therefore often not show up in instrument
budgets.
•
The time line of the project from section 9.2 is consistent with the current budget.
It is beyond the scope of this Small Study to provide an accurate cost breakdown for
stand-alone modules of MIDIR (e.g. for imager, medium and high resolution
spectrograph). However, one can already get an idea of the cost division between the
various components (imager/spectrometer) from Table 9-1.
For an individual
spectrometer mode (medium or high resolution) the breakdown is less obvious; the
dominant fraction of the cost is common for both modes: pre-optics, detector array,
cryostat, spectrometer software, etc. A rough guess is that dropping the HR mode would
result in a cost reduction in hardware of 3 MEuros, and in manpower of about 30 manyears. Dropping the high-resolution mode would result in a cost reduction of about 15%
(while a substantial fraction of the most innovative and competitive MIDIR science will
be lost). The cost reduction will be marginally smaller when dropping the medium
resolution mode.
Page 182 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
Table 9-1 Approximate cost estimate for the MIDIR project (continued on the following
pages).
COST ESTIMATES MIDIR
ITEM
No. of items
Cost/item
COST
(kEur)
TOTAL
A. HARDWARE
COMMON HARDWARE
Calibration unit
Active flexure control
Pre-optics
Pre-optics mechanisms
Pre-optics structure
Cryostat window
Cryostats spectrometers
Central cryostat/support structure
Vacuum equipm.
Heat shield
Cryo-cooler systems
Temperature control, sensors, heaters
Thermal links
Cabling, connectors
Readout electronics
Control electronics
Handling equipment
Spare parts (10% of total HW excluding DM)
9
25
1
75
3
1
100
150
4
12
50
60
24
25
200
225
50
75
30
75
300
150
150
200
720
100
100
100
600
250
100
343
3768
IMAGER
Collimator optics (TMA)
Camera optics (TMA)
Filters
Grisms
Cryomechanism
Cryomechanics
Harness, cabling, connectors
Models, prototypes
Auxiliary tools, handling equipm., test equipm.
Spare parts (10% of total HW excl. detectors)
Detector arrays LM
Detector arrays NQ
1
2
30
3
3
200
200
15
30
75
4
4
400
320
200
400
450
90
225
150
150
100
150
192
1600
1280
4987
SPECTROMETER
Dichroic beamsplitters
IFU's
MR-Collimator-Cameras
HR-Collimator-Cameras Cross-dispersion
HR-Collimator-Cameras Main-dispersion
Gratings MR
Gratings HR cross dispersion
Gratings HR main dispersion
CdTe lenses
Grating tilt mechanisms
Mode (MR/HR) switch mechanisms
Cryomechanics
Harness, cabling, connectors
Models, prototypes
Auxiliary tools, handling equipm., test equipm.
Spare parts (10% of total HW excl. detectors)
Detector arrays LM
Detector arrays NQ
2
3
3
3
3
3
3
3
6
9
3
3
15
150
150
150
200
50
50
150
15
75
75
100
4
6
400
320
30
450
450
450
600
150
150
450
90
675
225
300
300
150
300
477
1600
1920
8767
TOTAL HARDWARE
General comments:
17521
Detector costs include supportive electronics, and engineering grade samples
Optical component costs include mechanical mounts and frames
Page 183 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
COST ESTIMATES MIDIR
No. of myrs
Sum
B. MANPOWER
Optical design
Common HW
Imager
Spectrometer
2
6
12
Mechanical design
Common HW
Imager
Spectrometer
6
6
24
Thermal design
Common HW
Imager
Spectrometer
11
6
12
Electronics design
Common HW (incl. AO)
Imager
Spectrometer
AIT + instrument characterisation
Common HW
Detectors
Imager
Spectrometer
Integrated instrum.
6
6
4
14
8
Instrument control software
Common HW
Imager
Spectrometer
10
3
6
20
36
29
8
4
8
20
38
19
Data flow/storage + on-line analysis sofware
12
Off-line data-analysis software
16
Project management
Systems engineering
Administrative support
QA
12
12
8
6
TOTAL MANPOWER
228
(myr)
600
200
TRAVEL
TRANSPORT+INSURANCE
(myr)
OVERALL PROJECT COST
228
(kEur)
18321
Page 184 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Cost/item
COST
(kEur)
COST ESTIMATES MIDAO
ITEM
No. of items
TOTAL
A. HARDWARE
COMMON HARDWARE
AO - WFS
AO - DM
AO - Other Optics
AO - Structure
AO - Tip/tilt mechanism
AO - Control Electronics & Computing
Spare parts (10% of total HW excluding DM)
1
1
300
1400
1
75
300
1400
250
100
75
250
98
TOTAL HARDWARE
General comments:
2473 kEur
Detector costs include supportive electronics, and engineering grade samples
Optical component costs include mechanical mounts and frames
Effort
(myr)
B. MANPOWER
Optical design
Mechanical design
Thermal design
Electronics design
AIT + instrument characterisation
Instrument control software
TOTAL
6
4
1
4
6
10
TOTAL MANPOWER
31 myr
9.2 TIME LINE
To be able to serve as first light instrument, the schedule of the design and realisation of
MIDIR should match the telescope schedule. The scheme in Table 9-2 shows the time
estimate needed from start to completion of MIDIR. The following assumptions underlie
this scheme
•
•
Financing and contractual issues do not impact the time line of the project
The telescope design is assumed to be sufficient mature half 2009 to have a fixed
ICD towards the instruments
• Before this date, the information from the telescope interface and the instrument
requirements is sufficiently fixed to start earlier in the preliminary design (half a
year)
• The PDR will start out from a well defined conceptual baseline for the instrument
• Duration preliminary design: 1.5 years (proper concept available)
• Duration critical design: 2 years
• Production and sub-assembly integration: 2 years (long lead items leading this
phase)
• Final integration, test, verification: 1 year
The limiting factor in the present schedule is the availability of a sufficiently mature
telescope interface.
In addition to the “normal phases” for developing astronomical instruments, the Point
Design Study (PDS) is imperative to guide the telescope ICD process and define the
instrument parameters to a sufficient level of detail to make a swift start with the
preliminary design. In addition, this activity ensures the proper start and control of the
necessary technology programme.
Page 185 of 204
Doc. No
Issue
Conceptual Design Study of MIDIR
ELT-TRE-LEI-11200-0001
1.0
Parallel to the main work, other concepts, like polarimetry, or other spectrograph
principles, can be studied. This work should finish well before the start of the preliminary
design phase in order to have a proper concept as starting point. Other parallel activities
are technology readiness studies performed apart from the main stream of the instrument
design, with proper phasing towards the various milestones in the process. These studies
comprise e.g.: cryogenic AO systems, thermal background subtraction principles, optical
component production methods, etc.
Table 9-2: Estimated timeline for the MIDIR project.
ID
Task Name
1
Telescope design
Short Studies
Point Design Study
Other concepts study
Critical technology demonstrators
Technology demonstrators
Technology validators
Telescope ICD and requirement review
MIDIR Preliminary design
PDR
MIDIR Critical design
CDR
Production-Assembly integration and tes
Assembly Readiness Review
MIDIR full integration
Instrument Readiness Review
MIDIR shipment, test and installation
Commisioning
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Telescope ICD and requirement review
PDR
CDR
Assembly Readiness Review
Instrument Readiness Review
9.3 BASELINE OVERVIEW AND RISK ITEMS
So far, the instrument has been described in its detail. The overall overview of the baseline
is presented in this section. Due to the nature of this stage of the project, science case still
evolving, telescope interface not defined, certain issues can not be well defined in the
baseline yet. However, future progress will settle these issues. A Point Design Study is
needed to define both the instrument baseline and the instrument in more detail.
Figure 9-1 shows a block diagram of the instrument. The current figure assumes the AO
system to be inside the cryostat at a currently not specified temperature. The light from the
telescope (almost any f-ratio < 16 possible) enters the cryostat via an entrance window,
passes relay optics including an SCAO system towards a selector that switches between
imager and spectrometer. An external calibration unit is provided to couple light into the
system in an early stage. A fast switch mirror combines the optical paths of telescope and
calibration unit. The imager starts with a field mask in the focal plane of the instrument
pre-optics, after which the light is collimated (F/10 beam) for filtering, low resolution
dispersion, masking the thermal background and separation into two channels: one for LM
band and one for NQ-band imaging.
Page 186 of 204
Conceptual Design Study of MIDIR
Calibration
optics
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Telescope
Entrance window
AO
Selector
WFS
Pre-optics/Cold stop/
Dichroic splitting
Field lim. mask
Collimator optics
?
IFU 1, 2, 3
Switch HR/MR
Dichroic/Switch mirror
?
Cold-stop
Pupil imager
Cold-stop
Grisms
Grisms
Filter
Filter
Camera optics
Camera optics
Detectors
Detectors
HR/MR
spectrometers
Pupil
Imager
Switch HR/MR
Detectors 1, 2, 3
Figure 9-1: A block diagram for MIDIR. The imager components are shown in some
detail, the spectrometers are combined in one block to reduce the complexity of the figure.
In the spectroscopy mode, the light is switched to the spectrometer pre-optics, where the
beam is collimated, masked (cold stop) and split into three waveband channels. In each
channel an IFU converts a “square” FOV into a long virtual slit that is offered to the
spectrometer. A switch here selects between medium and high resolution. After the
spectrometers, the light is coupled into a focal plane array that is dedicated for each
spectrometer channel.
General issues, like pupil cameras and internal metrology systems, are not yet
incorporated in the design. Options like parallel observing modes in the instrument are
still subject to further detailing.
Table 9-3 presents the baseline together with an overview, its relation to the requirements,
alternative approaches and risk items connected to the design.
Page 187 of 204
Table 9-3: Hardware components of MIDIR.
Hardware
Baseline
Considerations
Options
Critical issues
Interface to telescope
Relay integrated in AO system,
F-ratio and pupil matching
DL performance, accept Fratio telescope
F/4.5 and F/16 interface (section
8.2)
None
AO
SCAO – cryogenic,
parameter in table 5.5
Performance
above
specification, based on DL for
20”x20”
Based on reduced requirement: 1
magnitude guide star gain, 4x less
actuators (2x linear), 2x decrease in
closed loop frequency
Cryogenic AO mirror
main
Cryogenic temperature or
ambient IR optimized (clean)
just
Two systems: one for F/16 -> F/10
and one for F/4.5 -> F/10 including
focal plane corrections
WFS (detector) close the observing band
Control, relaxed with other ELT, but still
XAO for current generations AO
Instrument relay
Integrated in AO relay
DL, offer F/10 with pupil far
upstream (section 6.1)
None
Calibration module
Section 6.6
Compliant to section 4.9
Vacuum window
Currently assumed to be a broad
band AR-coated CdTe window.
High efficiency band pass
between 3 and 27 μm
Have several windows on a window
exchange mechanism
Slower telescope F-ratio implies larger
windows. Homogeneity and throughput
might prove difficult.
Common pre-optics
Section 6.2
OK
Not critical
None
Imager
Section 6.3, comprises F/10.3
camera for 3.5<λ<5.5µm and
F/8.6
for
7<λ<20(27)µm,
grisms and filters, FOV =
40”x40”
Imager with 30 filters and low
resolution spectroscopy, long
slit in first focal plane in
instrument
Dichroic or mirror switch between
two arms
- Filters and dichroics efficiency and
performance
Still requires check on source intensities
needed and typical gas cell densities for
R=50000
- Mirrors: shape and surface finish for
λ~3.5µm
- Grisms homogeneity & coating
Dichroic Switchyard
spectrometer
Two mirrors/filters
6.2.4)
(section
Performance close to ok apart
from overall efficiency
Plain switching and blocking filters,
no parallel observing any more
- Filters and dichroics efficiency and
performance
Conceptual Design Study of MIDIR
IFU – LM
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Image slicer with FOV:
0”.8x0”.8, input matched to
dichroic chain, output to
spectrometer
requirements,
currently telecentric beam with
10<F/#<20, section 6.2
FOV should expand to at least
1”x1”.
No other technologies yet, fibre
based system not yet feasible
- Image quality of IFUs for these large
fields,
IFU-N
Image slicer with FOV:
1”.3x1”.3, see section 6.2
OK
See IFU-LM
Image quality of IFU, interface IFU
spectrometers without intermediate optics
IFU-Q
Image slicer with FOV:
1”.8x1”.9, see section 6.2
OK,
intermediate
optics
currently needed for F-ratio
matching
See IFU-LM
None
HR spectr. LM
Section 6.4, FOV 0”.8x0”.8,
Low order echelle grating +
filtering, Resolution R=50000
@ λ=5.1 µm
Too small FOV!
Cross dispersed system
• Grating quality
- large field IFU for LM,
- surface roughness slicer for λ~3.5 µm
• Size optics
• Quality of optics (shape and surface)
• Stability optics
HR spectr. N
Section
6.4,
X-dispersion
Echelle spectrograph, FOV
1”.3x1”.3, R=50000 @ λ=10.52
µm, coupled double pass TMAs
OK, large physical size, pupil
matching by field lenses,
Optical quality ok, F-ratio
coupling IFU might be
difficult (transfer optics?)
Different combinations of TMAs.
Working, but currently at limit.
A reduction in F-ratio leading to
oversampling and reduced FOV
might solve problem
• Grating quality and sizes
• Pupil matching
• Size optics
• Stability optics
• Fast camera F/2.2 including the 50%
oversizing (without oversizing still F/3.3)
HR spectr. Q
Section
6.4,
X-dispersion
Echelle spectrograph, FOV :
1”.8x1”.9, R=25000 @ λ=19.4
µm
Fast camera might be not
feasible. F-ratio coupling IFU
not directly possible.
MR spectr. LM
Section 6.4. First order grating
spectrograph
FOV too limited
Pupil size and overall dimensions
Quality of optics (shape and surface)
MR spectr. N
Section 6.4.
OK
Pupil size and overall dimensions
MR spectr. Q
Section 6.4
OK
Page 189 of 204
Conceptual Design Study of MIDIR
Detectors
LM -> 2kx2k HgCdTe Hawaii2RG arrays, pixel pitch 18 µm,
section 6.7, NQ -> Aquarius
1kx1k, Si:As array, pixel pitch
30 µm
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
OK, AR coating?? Well-depth
Various pixel sizes for the LM
detectors
• Closely buttable arrays
AO WFS
K-band detector
Mechanical structure
Stiff external structure, cold
benches currently Al, similar to
mirrors, stiffness reached by
active control of critical
components
OK
Other materials, needs to be studied,
mass a driver
Cryostats
Stainless steel
OK
Al or others materials
Vacuum system
Integrated in instrument
To be detailed later
Coolers
Electronics
• Pixel size preferably larger than 30 µm for
N and Q-band
To be worked on later
Weight
Not critical
• Pulse Tube
• Not all orientation
• Gifford McMahon
• Vibrations
• Helium Liquefier
• Weight, complexity
Not critical
Software
Page 190 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Table 9-4: Observational issues of MIDIR
Observational issue
Baseline
Considerations
Options
Critical issues
Field rotation
Field derotation by software
(section 8.4)
No degradation in Strehl ratio
anywhere in the FOV within
DIT
Mechanical
derotator,
by
instrument, detector or in the optical
chain
Optical chain option requires sufficient long
optical path
Atmospheric
Dispersion
No ADC in system
Moist in air can cause
problems. Close to Zenith with
dry air, it should be ok
Control at site selection
Atmospheric dispersion characterisation in
MIR regime for potential telescope sites
Water
fluctuations
vapour
Dry site selection
Atmospheric dispersion measurements at
various realistic moisture levels
Open four options to be
implemented or studied, normal
chopping/nodding scheme does
not work any more (see section
4.3 and 6.8)
Critical
AO
Good AO performance possible
for limited sky coverage for
λ<7µm
Too limited sky coverage
AO
WFS on other wavelength than
science target by many octaves.
Scaling possible?
Thermal background
• Focal plane chopping
• Pupil plane chopping
• Dicke Switching
• Nodding/Dithering
Performance of all MID-IR instruments
depend critical on thermal background
suppression. Various options need to be
studied. Detector response couples strongly
with background scheme.
Laser guide star?
Impact of laser guide star on MID-IR AO?
WFS at LM or even N band?
Impact of water vapour fluctuations on AO
Page 191 of 204
10 Conclusions and Outlook
We have presented a “Small” concept study of a mid-IR instrument for the E-ELT. Our
study has shown that exciting science cases for such an instrument exist, and that most of
them can only or better be done with MIDIR than with any other ground- or space-based
instrument.
Although MIDIR does not depend on fundamentally new technologies, certain
technologies need to be further developed, and additional design and operational aspects
need to be investigated. In general these items can be grouped in two categories, one for
which solutions are in principle known but where the details need to be worked out, and
one where the best approach or even the necessity is not yet sufficiently clear. Examples
are:
1. Items for which the details need further work:
•
Optical and opto-mechanical designs
•
Mechanical stability, actuation and system metrology
•
Cryogenic concept and power needs
•
Accurate estimates of mass, volume, stability
•
Operational issues like data rates, handling, and pipeline processing
•
Improved and expanded science cases
•
A comprehensive technical risk analysis
•
A performance simulator
•
Calibration schemes and provisions
2. Areas which require more study before the best approach becomes clear:
•
The impact of water vapour fluctuations on the image quality
•
The need for atmospheric dispersion correction
•
The best chopping scheme and its implementation
•
Operating a cryogenic AO system
•
The scientific need for polarimetry and it possible technical implementation
•
Manufacturing of large format Echelle gratings
•
The optimum cryostat window exchange mechanism
•
Manufacturing of high quality mid-IR filters and dichroics
•
The optimum IFU field of view and field geometry at LM bands
•
Parallel operating modes and their impact on data flows and instrument control
•
Detailed trade-off studies between science capabilities, instrument complexity,
schedule and cost.
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
This concept study represents the first important step toward a real mid-IR ELT
instrument. It is clear from the above list that much more work beyond this “Small Study”
is needed to cover all the relevant aspects for such an instrument. A comprehensive
funding and management structure is necessary to successfully support such a complex
project in the future. However, the current study has already shown that MIDIR is
scientifically attractive and technically feasible, and that the E-ELT would be the right
platform to advance mid-IR astronomy in the 21st century.
Page 193 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Annex A: Noethe 2003
Page 194 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 195 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Annex B: R. Siebenmorgen & H.U. Käufl 2006
Page 196 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 197 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 198 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 199 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 200 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 201 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 202 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 203 of 204
Conceptual Design Study of MIDIR
Doc. No
Issue
ELT-TRE-LEI-11200-0001
1.0
Page 204 of 204